EPIDEMIOLOGY OF NATURAL TRANSMISSION
OF BOVINE LEUKEMIA VIRUS INFECTION
MARK CY THURMOND
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL
FULFILLMENT OF THE REQUIREMENTS FOR
THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
Mark Cy Thurmond
Technical assistance with agar-gel immunodiffusion and
animal sampling was provided by Ms. J. Hennemann, Ms. J.
Ring, Mr. C. Maden, Mr. T. O'Donnell, Mr. A. Green, and
Mr. J. Lindsey. Most data management and computer program-
ming were performed by Ms. J. Galvez and Mr. D. Puhr. Edi-
torial assistance was provided by Dr. M. Burridge, Dr. P.
Nicoletti, and Dr. C. Wilcox. The typist was Ms. B.
Smerage. The valuable support and discussions offered by
the following people are gratefully acknowledged: Dr. M.
Burridge, Dr. R. Carter, Dr. M. Drost, Dr. C. Franti
(University of California/Davis), Dr. J. Gaskin, Dr. R.
Kahrs, Dr. J. Miller (USDA, Ames, IA), Dr. K. Portier,
Dr. O. Straub (West Germany), Dr. M. Van Der Maaten (USDA,
Ames, IA), Dr. S. Walter (Yale University), and Dr. C.
Wilcox. Additional thanks is extended to Dr. R. Carter,
Dr. K. Portier, and Dr. C. Wilcox for their interest and
support in statistical designs and analyses. Financial aid
was provided by United States Department of Agriculture
cooperative agreement 58-519B-0-872 and by the Wetterburg
Foundation of Newark, New Jersey.
TABLE OF CONTENTS
ACKNOWLEDGEMENTS . .
ABSTRACT . .
I INTRODUCTION . .
II LITERATURE REVIEW . .
Bovine Leukemia Virus Infection .
Bovine Leukemia Virus. .
Pathogenesis . .
Transmission Via Excretions and
Secretions . .
Serology . .
Factors Examined for Associations with
Bovine Leukemia Virus Infection .
Transmission of Bovine Leukemia Virus
Infection . .
Infection in Other Domestic Animals. .
Reviews . .
GENERAL MATERIALS AND METHODS .
Population Studied . .
Location and Climate . .
Management Practices . .
Sampling Design . .
Demographic Information .
Serology . .
Other Species Examined .
Diseases or Conditions Observed. .
Data Collection and Computer Programs.
IN UTERO TRANSMISSION OF BOVINE LEUKEMIA
VIRUS . .
Introduction . .
Materials and Methods .
Results . .
Discussion . .
V DECAY OF COLOSTRAL ANTIBODIES TO BOVINE
LEUKEMIA VIRUS . 41
Introduction . .. 41
Materials and Methods. . .. 41
Results . .. 44
Discussion . .. 45
VI AGE-SPECIFIC RATES OF DETECTION OF BOVINE
LEUKEMIA VIRUS INFECTION . 53
Introduction . 53
Materials and Methods . ... 54
Results. . .. 58
Discussion . . 60
VII SEASONAL PATTERNS OF RATES OF BOVINE
LEUKEMIA VIRUS INFECTION .. 83
Introduction . 83
Materials and Methods. . .. 84
Results. . .. 86
Discussion . ... 87
VIII SPATIAL PATTERNS OF BOVINE LEUKEMIA VIRUS
INFECTION. . . 98
Introduction . ... 98
Materials and Methods. . .. 99
Results. . . 104
Discussion . . 105
IX IATROGENIC TRANSMISSION OF BOVINE LEUKEMIA
VIRUS INFECTION. . 118
Introduction . .. 118
Materials and Methods. . .. .118
Results. . .. .120
Discussion . .. 120
X SUMMARY . . 129
A PLAT OF THE UNIVERSITY OF FLORIDA DAIRY
RESEARCH UNIT. . 134
B AVERAGE MONTHLY HIGH AND LOW TEMPERATURES
AND RAINFALL BETWEEN JULY 1, 1979, AND
SEPTEMBER 30, 1981, FOR GAINESVILLE,
FLORIDA. . . 135
C MONTHLY FREQUENCIES OF CALVES BORN ALIVE
BETWEEN JULY 1, 1979, AND JUNE 30, 1981,
AT THE UNIVERSITY OF FLORIDA DAIRY
RESEARCH UNIT. . .. .136
D LOCATION SITES AT THE DAIRY RESEARCH UNIT 137
E PRECIPITATION LINES OF AGAR-GEL IMMUNO-
DIFFUSION. . .. .144
LIST OF REFERENCES. . .. .145
BIOGRAPHICAL SKETCH . ... .166
Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
EPIDEMIOLOGY OF NATURAL TRANSMISSION
OF BOVINE LEUKEMIA VIRUS INFECTION
Mark Cy Thurmond
Chairman: Michael Burridge
Major Department: Animal Science
A 27-month study examined 473 dairy cattle for associa-
tions between bovine leukemia virus (BLV) infection and host
and environmental factors. Cattle sera were tested at
monthly intervals for BLV antibodies by agar-gel immuno-
diffusion using the glycoprotein-51 antigen. A model of BLV
colostral antibody decay in 130 calves predicted infection
in calves less than six months of age and estimated anti-
body half-life to be 27.1 1.2 days. Colostral antibody
decay did not differ between BLV-infected and noninfected
calves for slope (p= 0.45) or intercept (p= 0.43). By 95
days of age, 50% of the calves had no detectable BLV
Of 125 calves born to BLV-infected cows and followed
for at least four months, eight (6.4%) had precolostral BLV
antibodies, as determined by radioimmunoassay using the
glycoprotein-51 antigen. In utero infection withBLV was
not associated with dam age (p= 0.86), dam parity (p =0.83),
breed (p= 0.66), sex (p= 0.11), or stage of gestation in
which the dam was infected (p= 0.50). Calves infected in
utero did not pose an increased risk of infection to calves
penned next to them (p = 0.61).
Prevalence rates of infection were 9%, 16%, and 63% at
6, 16, and 27 months of age, respectively. Age-specific
rates of infection were not associated with dam age (p =
0.79), dam parity (p= 0.75), dam BLV-status (p= 0.46), breed
(p= 0.86), or BLV-status of colostrum consumed (p= 0.50).
An algorithm was described which allocated probabili-
ties of infection to locations occupied by an animal prior
to detection of infection. Small calf pastures were associ-
ated with less infection than was the calf barn (p < 0.05).
No less infection was associated with individual outdoor
calf pens compared to contiguous indoor pens (p> 0.05).
Risk of infection associated with the dry herd was five
times that for heifer pastures (p< 0.0001) and accounted
24 infections per 100 heifers per 100 days at risk.
Vaccination for infectious diseases was not associ-
ated with increased BLV infection (p= 0.33). Infection
rates were not associated with month of birth (p= 0.24) or
with season of potential arthropod vectors (p= 0.20).
Heifer infection was likely to occur in late winter or
spring (p = 0.01).
Bovine leukemia virus (BLV) has been shown to be the
causative agent of enzootic bovine leukosis, a neoplastic
disease of cattle (Callahan et al., 1976; Kettmann et al.,
1976; Miller et al., 1969). Bovine leukosis is believed to
have spread to western European countries from the Baltic
region during World War I (Bendixen, 1965). Following
World War II, efforts were undertaken in some European
countries to reduce the tumor incidence rate through hemato-
logic examinations for persistent lymphocytosis, a phenomenon
associated with bovine lymphosarcoma (Bendixen, 1965).
After discovery of BLV, and subsequent development and use
of serologic methods for mass screening (Hoff-Jorgensen et
al., 1978; Miller et al., 1969; Miller and Van Der Maaten,
1976a; Onuma et al., 1975; Schmidt et al., 1978), eradica-
tion of enzootic bovine leukosis progressed rapidly (Bause
et al., 1978; Mammerickx et al., 1978a; Straub, 1978b).
In order to preserve gains made in these programs,
restrictions were placed on BLV-seropositive cattle and on
semen entering countries either free from BLV or with BLV
control programs (Miller, 1980). Such restrictions have
placed an economic burden on the cattle export markets of
the United States (Mix, 1979). Because of the high genetic
quality of American cattle, eradication or control of BLV
infection using European methods of test and slaughter
would not be a pragmatic alternative for the American pro-
ducer. Interest, therefore, has focused on prevention of
transmission and test and segregation within a herd (Miller
and Van Der Maaten, 1978a).
A prerequisite to control of BLV in a herd is a clear
understanding of natural transmission of infection from
fetal life to adulthood or to the age at which heifers
would move to export markets. Several constraints make the
study of natural transmission patterns difficult and may
explain the lack of reports of long-term prospective studies
in the literature. A major obstacle is the necessity for a
large sample of animals to be tested at close intervals
over a long time period. At the same time changes in man-
agement or environmental factors must be recorded.
The device used to measure infection must be sensitive,
specific, inexpensive, simple, and meet specifications of
other programs. This necessitates the use of agar-gel im-
munodiffusion because it fulfills the above conditions
(Miller, 1980). Definition of infection by a serologic
test, however, has important limitations. For instance,
discrimination has not been made between colostral anti-
bodies and infection-induced antibodies in calves less
than six to seven months of age which consumed colostrum
from a BLV-infected cow (Ferrer et al., 1977b). Another
problem in defining infection is that seroconversion may
lag behind BLV infection by as much as two to three months
(Mammerickx et al., 1980; Straub, 1978b; Van Der Maaten and
Miller, 1978b, 1978c).
These constraints are not unique to the study of the
epidemiology of BLV transmission. It is important, there-
fore, that designs for the study and eventual control of
BLV be generally applicable to the study of other diseases
Examination of risks of BLV infection in a large cattle
population over a 27-month period is presented here as a
logical progression from the fetal environment to adult-
hood. Detection of BLV infection based on serologic cri-
teria is used as a proxy for infection. The intent is not
to attempt statements about specific routes of BLV infec-
tion, but to describe temporal and spatial patterns of
natural infection observed in animals studied. Furthermore,
factors possibly associated with deviations in those pat-
terns will be examined using existing analytic methods in
a framework applicable to other diseases. In addition, new
techniques are presented which improve the efficiency of a
Bovine Leukemia Virus Infection
Manifestations of infection with BLV vary from no
signs to persistent lymphocytosis or to lymphosarcoma
(Abramova et al., 1974; Grimshaw et al., 1979; Kenyon and
Piper, 1977; Kumar et al., 1978; Sorenson, 1979; Stober,
1968). Clinical signs of tumor involvement usually are
seen in cattle over five years of age and are referable to
the organ system involved (Abramova et al., 1974, Grimshaw
et al., 1979; Sorenson, 1979; Stober, 1968).
Sporadic bovine leukosis (i.e., juvenile, thymic, or
cutaneous leukosis) is not associated with BLV infection
(Bundza et al., 1980; Chander et al., 1977; Onuma, 1978;
Onuma et al., 1979; Richards et al., 1981; Straub and
In the absence of tumor involvement, BLV-infected
animals do not appear to suffer production losses (Langston
et al., 1978).
Bovine leukemia virus infection is a ubiquitous infection
throughout the world (Burny et al., 1980; Burridge et al.,
1981). A survey of cattle in the state of Florida recently
estimated the infection rate of BLV in dairy cattle to be
48% (Burridge et al., 1981).
Bovine Leukemia Virus
Discovery of BLV was made following phytohemagglutinin-
stimulation of lymphocytes from cattle with lymphosarcoma
(Miller et al., 1969). The virus is classified as a single-
stranded RNA retrovirus (Burny et al., 1980). It is spheri-
cal in shape with a diameter of 60-125 nm (Calafat et al.,
1974; Calafat and Ressang, 1977a, 1977b; Dutta et al.,
1970; Miller et al., 1969).
Several viral proteins have been described. There are
at least two glycoproteins, gp-30 (Dietzschold et al.,
1978) and gp-51 (Onuma et al., 1975), which constitute the
outer shell of BLV (Burny et al., 1980; Devare and
Stephenson, 1977; Driscoll et al., 1977). An ether-
resistant protein constitutes the internal or core anti-
gen, known as p-24 (Gilden et al., 1975; McDonald and
Ferrer, 1976; Miller and Olson, 1972). The BLV genome
codes for a reverse transcriptase which has a unique re-
quirement for Mg (Dietzschold et al., 1978; Gilden et
al., 1975; Graves et al., 1977; Kettmann et al., 1976).
Six mutant strains of BLV have been investigated recently
(Couez et al., 1981; Kettmann et al., 1981).
Several investigators have identified BLV as a C-type
virus (Burny et al., 1980; Dutta et al., 1970; Ferrer et al.,
1971; Kawakami et al., 1970; Mussgay et al., 1977; Stock
and Ferrer, 1972; Van Der Maaten et al., 1974; Weiland
and Ueberschar, 1976). Others have been reluctant to de-
scribe BLV as a B- or C-type virus since immature viruses
are rarely found outside the cell (Calafat et al., 1974;
Calafat and Ressang, 1977a, 1977b; Dekegel et al., 1977).
Comparisons of BLV with other retroviruses or onco-
viruses by molecular hybridization have shown that BLV is
biochemically distinct from Friend mouse leukemia virus and
visna maedi virus (Kaaden et al., 1977), Rauscher leukemia
virus (Kettmann et al., 1975, 1976), simian sarcoma (wooly
monkey) virus, murine sarcoma virus, feline sarcoma virus,
and feline leukemia virus (Kettmann et al., 1975, 1977).
Other studies have demonstrated a lack of cross-reactivity
between proteins of BLV and Mason Pfizer monkey virus
(McDonald and Ferrer, 1976; McDonald et al, 1976), and
bovine syncytia virus (McDonald et al., 1976).
Replication of BLV occurs mainly in B-lymphocytes
(Kenyon and Piper, 1977; Paul et al., 1977), but an associa-
tion with T-lymphocytes also has been reported (Takashima
et al., 1977). Further support for B-cell involvement was
found in the expansion of the B-cell population in BLV-
infected cattle (Kenyon and Piper, 1977; Kumar et al.,
Following subcutaneous inoculation of leukocytes from
a BLV-infected steer, BLV was isolated from the spleen
after eight days, from leukocytes after 14 days, and oc-
casionally from prescapular lymph nodes thereafter (Van Der
Maaten and Miller, 1978b). In that study, BLV could be
isolated from lymphocytes two to three weeks before a de-
tectable serologic response, and virus was not isolated
from the thymus. As few as 2500 washed lymphocytes from an
infected steer have been able to transmit BLV infection to
susceptible calves (Van Der Maaten and Miller, 1978c).
Integration of BLV in the Host Cell Genome
Results of studies using BLV-specific DNA probes sug-
gest that BLV is an exogenous virus (Callahan et al., 1976;
Deschamps et al., 1981; Kettmann et al., 1976, 1978a, 1978b,
1979a; Kukaine et al., 1979). The DNA from lymphocytes of
BLV-infected cattle has viral sequences that cannot be
identified in the DNA from lymphocytes of noninfected cat-
tle (Callahan et al., 1976; Kettman et al., 1976), or in
normal cell DNA from BLV-infected cows (Kettmann et al.,
The BLV provirus is integrated in several sites of
the lymphocyte DNA in cattle with persistent lymphocytosis,
but in only one or a few sites in the DNA of cells of
lymph node tumors (Kettmann et al., 1979a, 1980a, 1980b).
Less than 5% of peripheral lymphocytes in asymptomatic,
infected cattle contain the provirus, whereas up to 33% of
the circulating lymphocytes in cattle with persistent
lymphocytosis contain the provirus (Kettmann et al., 1980b).
Seroconversion following infection with BLV has been
found to occur between two and seven weeks in cattle ex-
perimentally inoculated (Mammerickx et al., 1980; Van Der
Maaten and Miller, 1978a, 1978b). Half of the animals in
these studies had seroconverted by five weeks postinocula-
tion. The seroconversion period was similar for sheep
experimentally inoculated by the intradermal route, oral
route, or by BLV-carrying tabanid flies (Gentile and Rutili,
1978; Mammerickx et al., 1980; Ohshima et al., 1981). The
seroconversion period for animals naturally infected is
considered to be less than three months (Straub, 1978b),
and the pattern of seroconversion is believed to be similar
to that for experimentally infected animals (Van Der Maaten
and Miller, 1978b).
Transmission Via Excretions and Secretions
It is well documented that BLV can be experimentally
transmitted to cattle via blood (Mammerickx et al., 1980;
Van Der Maaten et al., 1981a) and lymphocytes (Miller and
Van Der Maaten, 1978b; Van Der Maaten and Miller, 1978a,
1978b) from infected animals. The virus has been demon-
strated in saliva but not in prostatic fluid or feces of
infected cattle (Ressang et al., 1980). The p-24 antigen
of BLV has been found'in urine of naturally infected
animals (Gupta and Ferrer, 1980).
Semen collected by manual massage from a BLV-infected
bull transmitted BLV infection to susceptible sheep (Lucas
et al., 1980). Another study failed to demonstrate BLV in
semen collected from BLV-infected bulls following normal
ejaculation (Miller and Van Der Maaten, 1979).
Several serologic tests for the detection of BLV have
been described for both the gp-51 and the p-24 antigen
(Burny et al., 1980). Agar-gel immunodiffusion using gp-51
has been recommended for use by member countries of the
European Economic Community (Kaaden and Stephenson, 1978),
because of its high sensitivity, simplicity, and low cost
(Miller, 1980). Recently a radioimmunoassay procedure
was described using gp-51 (Bex et al., 1979). This test
may be the most sensitive one presently available (Miller
et al., 1981).
Factors Examined for Associations with Bovine
Leukemia Virus Infection
A genetic predisposition to enzootic bovine leukosis
and bovine lymphosarcoma was suspected before discovery of
BLV. Leukosis was observed more frequently in daughters
of affected cows than in daughters of unaffected cows
(Bendixen, 1965; Larson et al., 1970). Pedigree studies of
lymphosarcoma found clustering of cases by sire and/or dam
families (Crowshaw et al., 1963; Cypess et al., 1974;
Marshak et al., 1962). It also was observed that herds
which were inbred experienced higher rates of leukosis and
lymphosarcoma than did noninbred herds (Abt, 1968; Laktionov
and Nakhmanson, 1972), but purebred herds were found to
have lower rates of leukosis than nonpurebred herds
(Anderson et al., 1971).
One study estimated the heritability of susceptibility
to BLV infection to be 0.44 0.22 (Burridge et al., 1979).
A study of lymphosarcoma, however, was not able to associate
the disease with serologically defined antigens controlled
by the BoLA-A locus (Takashima and Olson, 1978).
Parental Infection with Bovine Leukemia Virus
The effect of BLV infection of the dam on subsequent
BLV infection in the offspring has been examined by several
investigators. In one report, dam status appeared to have
a significant influence on progeny infection (Baumgartener
et al., 1978), but the authors felt that such an effect may
have been due to high prevalence rates in some herds. In a
longitudinal study, no association was found between dam
status and age at which progeny became infected (Olson et
al., 1978). Reports on cross-sectional studies concluded
that presence of BLV antibodies in the dam was not associ-
ated with subsequent progeny infection (Hofirek, 1980;
In a large study of progeny from BLV-infected, AI bulls,
offspring from infected sires did not have as high a rate of
subsequent BLV infection as did those from noninfected sires
(Baumgartener et al., 1978).
Few studies have examined for associations between
BLV infection and sex. Reports in which sex effects were
studied suggested no difference in infection rates between
males and females (Baumgartener et al., 1975; Evermann et
It has been reported that many different breeds are
susceptible to BLV infection (Burridge et al., 1981; Marin
et al., 1978). Analysis of data from a survey of Florida
cattle suggested that Jerseys had a higher infection rate
than Holsteins (Burridge et al., 1981). However, a study
within a Florida dairy herd indicated no difference existed
between rates of infections for Jerseys and Holsteins
(Burridge et al., 1979).
Age-specific prevalence rates of BLV infection have
been shown to follow a characteristic sigmoidal curve.
Rates increased linearly from one to four years of age,
after which they plateaued (Burridge et al., 1979, 1981;
Chander et al., 1978; Evermann et al., 1980; Ferrer et al.,
1976; Hofirek, 1980; Mammerickx et al., 1978a, 1978b;
Marin et al., 1978; Olson et al., 1973; Piper et al., 1979).
Peak prevalence rates of infection were observed at four
years of age in beef cattle and at more than nine years of
age in dairy cattle (Burridge et al., 1981). Ages at which
a sharp, linear increase in rates occurred varied from study
to study and from herd to herd. In some herds rates of
infection began to level off at two to four years of age
(Burridge et al., 1979; Olson et al., 1973), while in other
herds rates reached a plateau at four to five years of age
(Chander et al., 1978; Hofirek, 1980; Mammerickx et al.,
1978a, 1978b; Marin et al., 1978).
A few studies have approached age-specific rates of
infection in a longitudinal design using birth cohorts.
Results of one of these investigations showed that animals
of similar ages experienced different rates of infection,
according to the birth cohort (Wilesmith et al., 1980).
In the other study, each new 12-month cohort entered the
herd with a lower prevalence rate than did the previous
cohort. Rates within a cohort did not appear to change as
cattle aged. It was suggested further that higher preval-
ence rates of infection observed in older animals in
cross-sectional studies represented high-rate cohorts
(Huber et al., 1981).
Transmission of Bovine Leukemia Virus Infection
Rates of natural in utero infection with BLV have been
reported to range from 3% to 25% (Ferrer et al., 1976,
1977a, 1977b; Piper et al., 1979). Stage of gestation dur-
ing which a dam is experimentally infected has not been
associated with the frequency of infection in progeny (Van
Der Maaten et al., 1981b).
Close physical contact between infected and susceptible
cattle is believed to be a prerequisite to BLV transmission
(Ferrer and Piper, 1981; Maas-Inderwiesen et al., 1978;
Miller and Van Der Maaten, 1978a; Wilesmith et al., 1980).
Newborn calves were more likely to develop leukosis when
placed in close contact with leukotic cows (Straub, 1971).
Infection rates increased during winter months in one herd
studied, suggesting transmission associated with indoor
housing conditions (Wilesmith et al., 1980). Limiting
physical contact by vacating a stall between animals or by
placing a single fence between animals appeared to retard
transmission of infection (Miller and Van Der Maaten, 1978a).
Bovine leukemia virus has been isolated from the mid-
gut of horseflies after feeding on a BLV-infected cow
(Bech-Nielsen et al., 1978). In an experimental study,
horsefly transmission of BLV infection to sheep was demon-
strated (Ohshima et al., 1981). It has been suggested
that high rates of infection observed in animals during
summer months support the hypothesis of vector-borne trans-
mission of BLV infection (Bech-Nielsen et al., 1978; Onuma
et al., 1980). Another study, however, observed higher
rates of infection during winter months (Wilesmith et al.,
Ixodes ricinus ticks have been suggested as a possible
explanation for geographic differences in rates of infec-
tion in Sweden (Hugoson and Brattstrom, 1980).
Intranasal instillation of BLV-infected lymphocytes
produced infection in one of two calves and an aerosol
exposure to BLV-culture fluids produced infection in two of
two calves (Van Der Maaten and Miller, 1978c). Both methods,
however, also exposed the oral cavity and the latter method
exposed the conjunctivae.
Bovine leukemia virus or BLV-like particles have been
identified in milk and colostrum of BLV-infected cows and
cows with lymphosarcoma (Dutcher et al., 1964; Jensen and
Schidlovsky, 1964; Miller and Van Der Maaten, 1979;
Schulze et al., 1966). Transmission of BLV by the oral
route has been demonstrated by inoculation of leukemic
blood into neonatal lambs (Mammerickx et al., 1976a), by
inoculation of lymphocyte cultures into colostrum-deprived
calves (Miller et al., 1972; Van Der Maaten and Miller,
1978c), and by feeding BLV-infected lymphocytes in colostrum
free of BLV antibodies (Van Der Maaten et al., 1981a).
Oral transmission has not been shown in old lambs (Hoss
and Olson, 1974) or in old calves (Van per Maaten and
Miller, 1978c). Infection did not occur when calves were
fed BLV-infected lymphocytes in colostrum containing BLV
antibodies (Van Der Maaten et al., 1981a). Some studies
concluded that considerable emphasis should be placed on
oral transmission, especially when bloody colostrum or
milk is fed (Parfanovich et al., 1978; Seger and Morgan,
1977). Results of other studies suggest that milk or
colostrum does not constitute a major vehicle for trans-
mission of BLV infection (Ferrer et al., 1976; Ferrer and
Piper, 1978, 1981; Piper et al., 1975, 1979). Pasteuriza-
tion of milk or colostrum for 30 seconds at 600C would
probably inactivate the virus (Baumgartener et al., 1976).
One study demonstrated the venereal transmission of
BLV infection after placing BLV-infected lymphocytes in the
cervical canal of susceptible cows (Van Der Maaten and
Intradermal inoculation of BLV has been shown to be a
viable means of transmission (Van Der Maaten and Miller,
1978c). Iatrogenic transmission of BLV following blood
sampling procedures has been suspected (Bause et al.,
1978; Maas-Inderwiesen et al., 1978; Wilesmith, 1979).
Premunization of cattle for babesiosis and other vaccina-
tion procedures also have been incriminated in BLV trans-
mission (Hugoson and Brattstom, 1980; Hugoson et al.,
1968; Marin et al., 1978; Stamatovic and Jonavic, 1968).
However, transmission of BLV to sheep following intradermal
Tb testing has not been demonstrated (Roberts et al., 1981).
Infection in Other Domestic Animals
The possibility for BLV infection in species other
than the bovine was first suggested when bovine leukosis
and later bovine lymphosarcoma were transmitted to sheep
(Olson et al., 1972; Wittmann and Urbaneck, 1969). Ex-
perimental infection with BLV has been confirmed in sheep
(Bansal and Singh, 1980; Bex et al., 1979; Gentile and
Rutili, 1978; Hoss and Olson, 1974; Mammerickx et al.,
1976a, 1976b, 1980, 1981; Olson and Baumgartener, 1978;
Van Der Maaten and Miller, 1976a), in goats (Hoss and
Olson, 1974; Mammerickx et al., 1981; Olson et al., 1981),
and in pigs (Mammerickx et al., 1981). Evidence exists
for experimental BLV infection in rabbits, but not in rats
(Bansal and Singh, 1980), and in guinea pigs (Lussier and
Pavilanis, 1969). Serologic evidence for BLV infection in
chimpanzees has been found (Van Der Maaten and Miller,
1976b). Passage of BLV through cattle, sheep, or goats
did not reduce the infectivity of the virus for cattle,
sheep, goats, or pigs (Mammerickx et al., 1981). Sheep
have not been found to shed BLV and thus may not be a
reservoir (Mammerickx et al., 1976b; Van Der Maaten and
Bovine leukosis, bovine lymphosarcoma, and bovine
leukemia virus infection have been extensively reviewed
(Bendixen, 1965; Burny et al., 1978, 1980; Ferrer, 1977,
1979, 1980a, 1980b; Ferrer et al., 1978, 1979; Hoff-
Jorgensen, 1977; House et al., 1975; Markson, 1979;
Mussgay and Kaaden, 1978; Olson, 1974; Olson et al.,
1970; Reed, 1981; Ruppanner and Paul, 1980; Tyler, 1978;
Van Der Maaten and Miller, 1975).
GENERAL MATERIALS AND METHODS
Cattle studied were those in the University of Florida
Dairy Research Unit (DRU) herd. This population was chosen
because it had been characterized previously by a BLV
prevalence rate of 75% (Burridge et al., 1979). Calves
born from July 1, 1979, through June 30, 1981, were fol-
lowed until death, sale, parturition, or September 30, 1981,
whichever occurred first.
Location and Climate
The DRU was located in north-central Florida (latitude
82030' west, longitude 29040' north). It occupied about 450
hectares of reclaimed pine flatwoods 20 km. northwest of the
University of Florida at Gainesville, Florida (APPENDIX A).
The climate was characterized by hot, humid summers and
cool, dry winters (APPENDIX B).
The DRU was a closed herd used as a research facility.
Two 18-month-old heifers, a Jersey (Ml01) and a Holstein
(M121), entered the herd, however, from a commercial dairy
(Bassett's Dairy, Montecello, FL) on May 5, 1980. Both were
negative for BLV antibodies when they entered the herd. The
Jersey had BLV antibodies by April 25, 1981, and the Hol-
stein remained serologically negative through September 26,
Approximately 200 purebred cows were milked, 1/3 of
which were Jerseys and 2/3 Holsteins. Cattle were allowed
to.graze, but it was necessary to supplement rations with
corn silage; corn, sorghum, ryegrass, or alfalfa greenchop;
Cattle were bred artificially using commercially avail-
able frozen semen. Some semen was used, however, from
young Jersey control sires raised at the DRU. Calving
season ranged from June through December (APPENDIX C).
Calves were born in small pastures (APPENDIX D, Fig. 1) and
remained with their dams for about 12 hours. They then
were placed in either contiguous, wire pens in the calf
barn (APPENDIX D, Fig. 2) or in individual outdoor pens
(APPENDIX D, Fig. 3). Once in pens, they were fasted for
12 to 24 hours, after which time they were fed either
pooled colostrum (obtained from cows one through three days
postpartum) or bulk-tank milk for the next two to three
days. Prior to weaning at about one month of age, calves
were fed bulk-tank milk once a day. Holsteins received
9 lbs. per day and Jerseys 7 lbs. per day. Calves also
were fed concentrates ad libitum from two or three days of
Electric dehorning, ear tagging, and tattooing were
performed during the first month of life. This usually was
not done in a serial fashion from calf to calf at one ses-
sion, but individually, one or two calves per day.
All heifers were vaccinated for brucellosis,a clostridi-
al diseases, and leptospirosis at four months of age
and for infectious bovine rhinotracheitis (IBR),d bovine
virus diarrhea (BVD),d and parainfluenza-3 (PI-3)d at 11
months of age. Brucellosis vaccination was administered
subcutaneously in the neck and other vaccinations intra-
muscularly in the gluteal area. Cattle were vaccinated in
groups using one or two common 18 gauge 1-1/2 inch needles
and a multidose syringe. The order in which cattle were
vaccinated was not recorded.
At the time of brucellosis vaccination, calves were
wormed eitherwith thiabendazolee or levamisole. The
specific type administered to each calf was not recorded.
aStrain 19, Colorado Serum Co., 4950 New York St., Denver, CO.
Siteguard, Clostridium chauvoei, septicum, haemolyticum,
novyii, sordellii, perfringens Type C and D. Jensen-
Salsbery, Kansas City, MO.
Novalep GHP, Jensen-Salsbery, Kansas City, MO.
RESBO 3, Norden Laboratories Inc., Lincoln, NE.
eOmnizole (paste), Merck Animal Health Division, Merck
& Co., Inc., Rahway, NJ.
Ripercol L, American Cyanamid Co., Princeton, NJ, or
Levasole, Pitman-Moore, Inc., Washington Crossing, NJ.
At two to three months of age, calves were moved from
indoor pens in the calf barn, or from nearby outdoor pens,
to paddocks (APPENDIXD, Fig. 4) or to small pastures (0.3
to 0.6 ha.) (APPENDIX D, Fig. 5). At about 10 months of
age, they were moved to larger pastures (2 to 3 ha.)
(APPENDIX D, Figs. 6 and 7). Cattle in each location were
of similar ages.
Heifers were bred between 13 and 16 months of age,
except during summer months when estrous behavior was less
apparent. At 40-45 days gestation, heifers were placed, for
the first time, with adult cattle in the dry herd. For
heifers under study, this began in early January, 1981.
Holstein heifers were removed from the dry herd on May 1,
1981, and placed in pastures holding only bred Holstein
heifers. Jersey heifers continued to be placed in the dry
herd and remained there until shortly before calving.
Most bull calves were sold a few days after birth, al-
though some were retained for veal studies or semen collec-
tion. Management of heifers was similar to that practiced
on most commercial dairies. Heifers usually were not used
in experimental studies until after freshening.
Fly control was practiced sporadically two or three
times during summer months. Lactating cows were hand-
dusted with a commercial insecticide,a and fly baitb was
aMarlate 50, E. I. DuPont de Nemoud and Co., Inc.,
Golden Malrin, Starbar Division of Zoecon Corp., 12200
Denton Dr., Dallas, TX.
scattered in the calf barn a few times during peak calf
density in the fall. Pastured heifers were not treated
for flies. Most flies observed in the calf barn were
Musca domestic feeding on decaying grain and soured milk.
Pastured cattle were bothered by large numbers of Haematobia
irritans and Stomoxys calcitrans and some tabanidae. No
ticks or lice were observed during the 27-month study.
Precolostral blood samples were drawn on nearly all
calves born alive between July 1, 1979, and September 30,
1980. To increase the chance of obtaining precolostral
samples, udder bags were placed on cows about one week
prior to calving (APPENDIX D, Fig. 1). This prevented the
calf from nursing until a blood sample could be drawn. In
order to reduce the period of colostrum deprivation, re-
search field personnel were notified by telephone as soon
as a cow calved. This procedure was instituted in December
1979, and calves born after this time were fed 1-2 liters
of colostrum from their own dam within one hour following
birth. After feeding, the navel was washed with a solution
of 3% Lugol's iodine. Through December 1979, calves were
bled precolostrally, at one week of age, and then at three-
month intervals. After December 1979, calves were bled
precolostrally, at two to three days of age, and then at
aNumbers C4591N-C4593N, Nasco, Fort Atkinson, WI.
monthly intervals beginning some time during the first month
of age. Calves born between September 30, 1980, and June 30,
1981, were not bled precolostrally, but were bled at monthly
intervals beginning sometime during the first month of age.
Calves born after June 30, 1981, were not included in the
study. Cows were bled at two, five, and eight months of
gestation, at parturition, and at one month postpartum.
All blood samples were collected from either the jugular or
intercoccygeal vein using a sterile 20 gauge 1-1/2 inch
needle and Vacutainer tube.b
The following information was obtained for each calf:
BLV status of dam, age of dam, parity of dam, breed, sex,
birth date, sale date, death date, or freshening date.
Locations and inclusive dates of occupation in pens and
pastures were recorded for all cattle as long as they were
followed. As soon as heifers began entering the dry herd,
all cattle in that group were sampled monthly to determine
the BLV status of animals exposed to heifers being studied.
No interventions were made in routine management of calves
or heifers except for colostrum feeding, navel washing, and
monthly mustering and bleeding.
aNo. 5746, Becton-Dickinson and Co., Rutherford, NJ.
bNo. 6512, Becton-Dickinson and Co., Rutherford, NJ.
Serum samples were tested for presence of antibodies
to BLV glycoprotein-51 antigen (gp-51)a using agar-gel im-
munodiffusion (AGID), as previously described (Burridge
et al., 1979). Gel plates were incubated at 24-27 C in an
humidified chamber for 48 hours before being read with the
aid of a direct light beam. Formation of precipitation
lines of identity with positive control sera indicated
presence of BLV antibodies (APPENDIX E). Serial two-fold
dilutions using phosphate buffered saline c were made of
all positive sera and then these dilutions were retested to
determine end-point titers.
All precolostral sera from calves lost to follow-up
before four months of age and negative on AGID were re-
tested using a radioimmunoassay (RIA) procedure with gp-51
(Bex et al., 1979).d Samples precipitating more than 15%
of labeled antigen were considered positive for BLV anti-
bodies. Persistence of antibodies or an increase in BLV
antibody titer was considered evidence of BLV infection.
Precolostral sera from calves with evidence of BLV infec-
tion at some later date were retested by RIA.
aAntigen supplied in the Leukassay-B kit, Pitman-Moore,
Inc., Washington Crossing, NJ, and by Dr. J. Miller, USDA,
bModel 651, American Optical Corp., Buffalo, NY.
cDulbecco's formula, Flow Laboratories, Inc., McLean, VA.
Performed by Dr. M. J. Schmerr, USDA, Ames, IA.
Other Species Examined
The DRU maintained between six and eight sheep and
goats during the 27-month study. These animals were kept
in a small pasture, separated from any cattle under study
by at least two fences. Tests for presence of antibodies
to BLV were negative for all sheep and goats in the fall of
1980 and 1981.
Several cats also inhabited the DRU during the study
period. Three of the tamer cats were bled in October 1981,
and no detectable BLV antibodies were found.
Diseases or Conditions Observed
During the 27-month study, lymphosarcoma was diagnosed
in two cows following postmortem examination. Both were
Jerseys, one six years old and the other five years old.
Although no health records were kept, nearly all neo-
natal calves experienced at least one episode of diarrhea
and bronchopneumonia, and several died. According to clini-
cal and pathological reports, a common cause of death,
particularly prior to December 1979, was septicemia result-
ing from omphalophlebitis and/or hypogammaglobulinemia.
Chronic dysentery was prevalent among calves two to six
months of age. This was probably due, at least in part, to
coccidia and trichostrongyles, as suggested by oocysts and
eggs found on fecal flotation. Many calves in this age
group, particularly bull calves in nutrition trials, were
quite cachectic. Heifers older than six months appeared
Data Collection and Computer Programs
Data were transcribed onto computer sheets from which
IBM cards were punched. Files were maintained on tape at
the Northeast Regional Data Center (NERDC), University of
Florida, Gainesville, Florida. Computer systems available
through NERDC for statistical analysis were the Statistical
Analysis System (SAS) version 79.5,a Biomedical Computer
Programs (BMDP), and the McGill University System for
Interacting Computing (MUSIC) version 4.1, 1978.
aSAS User's Guide, SAS Institute Inc., Raleigh, NC.
bBMDP, P-series, University of California Press, Berkeley,
IN UTERO TRANSMISSION OF BOVINE LEUKEMIA VIRUS
Since BLV is currently considered an exogenous virus
(Callahan et al., 1976; Kettmann et al., 1976, 1978a,
1978b, 1979a, 1979b; Kukaine et al., 1979) and there is
no evidence for natural transmission of BLV via semen
(Baumgartener et al., 1978; Miller and Van Der Maaten,
1979; Ressang et al., 1980), prenatal infection is most
likely a consequence of in utero transmission. Suggested
rates for in utero transmission of BLV infection have ranged
from 3% to 25% (Ferrer et al., 1976, 1977a, 1977b; Piper
et al., 1979). Highest rates were observed, however, in
an inbred herd in which selection for bovine leukemia had
been practiced (Piper et al., 1979). The BLV status of the
dams was not stated in some reports (Ferrer et al., 1976,
1977a, 1977b); therefore, rates could have been higher
had they been stated correctly in terms of calves born to
It is not known what factors, if any, predispose a
fetus to infection with BLV. Stage of gestation in which
the dam was infected has not been associated with fetal
infection (Van Der Maaten et al., 1981b). If predisposing
factors are present, calves and/or dams with the related
characteristic could be segregated, thus reducing post-
natal transmission. Otherwise, efforts could be focused on
other aspects of transmission control.
The purpose of this chapter is to describe observed
rates of natural precolostral antibodies to BLV in a large
sample of random-bred calves and to examine for associa-
tions between those rates and certain characteristics of
the calves and their dams.
Materials and Methods
Calves examined were those born to BLV-infected dams
and bled precolostrally. A BLV-infected dam was defined as
one which had BLV antibodies up to one month postpartum.
Calves were divided into two groups: (1) those post-
colostrally positive for BLV antibodies and followed for
four months or more, or postcolostrally negative for BLV
antibodies and followed for one month or more, and (2) those
with postcolostral BLV antibodies and not followed for four
months. A positive precolostral sample was considered
invalid, if, after following the calf for four months or
more, there was no evidence for BLV infection. Evidence
for BLV infection was based on a rising antibody titer
during the first six months of age, on observation of titers
outside the 95% prediction level of decay of BLV colostral
antibodies (CHAPTER V), or on persistence of BLV anti-
bodies beyond six months of age.
Rates of precolostral antibodies in both groups of
calves were examined for associations with breed and sex of
calf. Analyses were performed using the FUNCAT program, a
procedure for analysis of categorical data (Grizzle et al.,
1969) offered by SAS. The model was presence (or absence)
of precolostral BLV antibodies = breed + sex + breed x sex.
The response function was the difference between the pro-
portion of calves with and the proportion without pre-
colostral antibodies to BLV, or, as indicated by SAS, 1 -1.
Differences in age and parity distributions for dams
of calves with precolostral antibodies and dams of calves
without precolostral antibodies were examined using the
NPAR1WAY program of SAS, a two-sample Wilcoxon Rank Sums
test for nonnormal distributions. Fisher's exact test,
from the STATPAK subsystem of MUSIC, was used to calculate
the probability of association between the stage of gesta-
tion of seroconversion to BLV by the dam and occurrence of
precolostral antibodies in calves followed four months or
Of 346 calves born during the period July 1, 1979,
through September 30, 1980, 280 calves were bled
precolostrally and, of those, 223 were from BLV-infected
dams. Because many calves were sold or were not followed
for more than a few months, only 125 calves were born from
BLV-positive cows, bled precolostrally, and followed for
four months or more (Table IV-1). Eight (6.4%) of these
calves had precolostral antibodies and antibodies persist-
ing beyond six months of age (Table IV-1, IV-2). In the
group of 223 calves, 18 (8.1%) had detectable precolostral
antibodies (Table IV-1).
Precolostral antibodies were detected with similar
frequency in Jersey and Holstein calves, both in the fol-
lowed group (p= 0.66) and for all calves (p= 0.80) (Table
IV-3). When all calves were examined, males showed a
significantly higher rate of precolostral antibodies than
did heifers (p=0.04). There was no such association in
the 125 calves followed (p= 0.11) (Table IV-3). No sig-
nificant interactions between breed and sex were detected
for either the group followed (p=0.91) or for all calves
(p =0.64) (Table IV-3).
No statistical differences were detected between dis-
tributions of age (p= 0.86) or parity (p=0.83) for dams
having calves with precolostral antibodies and for dams
having calves without detectable precolostral antibodies
(Table IV-4). Complete serologic records were available
for 76 cows seroconverting before the second month of ges-
tation and for 14 cows seroconverting after the second
month. Those seroconverting after the first trimester of
gestation showed no more tendency to have a calf with pre-
colostral antibodies to BLV than did cows seroconverting
before the first trimester (exact p= 0.50) (Table IV-5).
The ruminant fetus may be particularly susceptible to
infection because (1) syndesmochorial placentation does not
allow transfer of maternal antibodies to the fetus; (2) the
fetal immune system is not fully functional; and (3) micro-
bial activity can take place in fetal cells (Osburn, 1981).
Since placentation restricts passage of antibodies to the
fetus, presence of fetal immunoglobulins is indicative of
infection by, or exposure to, an antigen or of placental
leakage of globulins (Brambell, 1970; Husband et al., 1972).
Although no estimates are available for frequency of
placental leaks, such events are believed to be rare
(Brambell, 1970). This has been confirmed by several
studies of in utero infection based on fetal infection and
fetal or precolostral serology (Braun et al., 1973; Dunne
et al., 1973; Fennestad and Borg-Petersen, 1962; Gibson
and Zemjanis, 1973; Horner et al., 1973; Kniazeff et al.,
1967; Osburn and Hoskins, 1971; Osburn et al., 1974;
Van Der Maaten et al., 1981b).
The main difficulty when studying in utero infection
as it occurs naturally in a population, using serologic
detection, is one of logistics. Efforts should be made to
sample the calf as soon after birth as possible and before
it nurses. For this reason, cows were fitted with udder
bags a week or two before their due-date. However, prior
to December 1979, research personnel were lax about udder
bag procedures and precolostral sampling. As a result,
some calves could have sucked BLV-positive cows without
udder bags. Precolostral samples could be partially veri-
fied, however, by following calves to determine if they
became serologically negative. For purposes of analysis,
two groups of calves were described based on confidence in
precolostral samples. A follow-up time of four months was
selected because 80% of calves consuming colostrum contain-
ing BLV antibodies were negative by 120 days of age (CHAPTER
V),and prediction of infected calves, therefore, could be
made with confidence.
Radioimmunoassay using gp-51 antigen has been shown
to be a highly sensitive test of BLV antibodies and may
detect infected animals up to 10 days sooner than AGID
(Miller et al., 1981). After screening with AGID, negative
sera were retested by RIA to identify calves infected in
late gestation. Additional calves identified as precolos-
trally seropositive by RIA may have consumed colostrum, as
previously mentioned. More confidence, therefore, should
be placed on the precolostral antibody rate of 6.4% than on
that of 8.1% because of these logistical problems.
Since reported in utero infection rates of 10% or
more may have counted calves born to negative dams, those
figures represent minimal rates for calves born to infected
dams. In addition, syncytium induction assay, virus neu-
tralization, and/or immunofluorescence assay were used in
those studies to detect BLV or BLV antibodies. These tests
may be less sensitive than RIA with gp-51 for detection of
BLV infection (Miller et al., 1981; Van Der Maaten and
Miller, 1977), and, therefore, in utero infection may have
been underestimated further.
The most likely explanation for the discrepancy between
rates observed in this study and those reported in other
studies rests in innate features of the populations ex-
amined. Higher rates (e.g., 14%, 18%, and 25%) were ob-
served in a Jersey herd in which animals had been inbred
and selected for leukemia (Piper et al., 1979). Such high
rates may suggest a weakening of placental integrity due to
inbreeding allowing for passage of maternal blood into the
fetus. In any case, it is unlikely that the higher rates
are representative of in utero infection in a typical dairy
A rate of 3% was observed in 37 calves from a multiple-
case herd (Ferrer et al., 1976). Again the BLV status of
dams was not reported, and, therefore, rates in calves from
infected cows were possibly higher. Furthermore, caution
should be taken in considering a rate based on only one in-
fected calf from a small sample of 37 calves.
Results of this study indicated that the risk of BLV
transmission did not differ between older, multiparous cows
and younger or primiparous cows. This suggests that
transmission is not a consequence of loss of placental
integrity associated with increasing physiologic age or
pregnancies. In addition, stage of gestation at maternal
infection did not appear to influence fetal infection rates,
in accordance with observations from experimental studies
(Van Der Maaten et al., 1981b).
Reports on studies of breed effects on BLV infection
rates indicate that many breeds are susceptible (Burridge
et al., 1981; Marin et al., 1978). One report concluded
that higher infection rates occurred in Jerseys (Burridge
et al., 1981), but these differences could have been at-
tributable to high rates in particular age cohorts (Huber
et al., 1981; Wilesmith et al., 1980). Furthermore, no
breed effects were found in a previous study at the DRU
(Burridge et al., 1979).
Although fetal sex showed some statistical association
with precolostral antibodies when both groups were combined,
this possibly resulted from less fastidious postnatal
sampling of bull calves. Colostrum may have been consumed
by more bulls than heifers prior to sampling because less
attention is usually paid to postnatal care of bulls.
From a practical standpoint, knowledge of dam age or
parity probably would not assist in control or prediction
of in utero BLV infection. Economic justification for
efforts to control in utero infection seems unlikely with
a rate as low as 6.4%. Even though control may not be
possible, awareness of in utero infection is essential
in planning postnatal management of BLV transmission.
( o 0
*-d C) rC-
rd (U) r1
6 rd S-
,U -) t -
r-o Cz (
0 () Q
04 Q) '-
CO *l C
'o 0 m in
T C N (N
m (N CN r-l
LO M m r-
(N- CN 0w
m or- No
m0 0 -- CO
CO 0 co
N 0 cy
En n 0-
M 0- (N
0 0 0
- -c- O.
5 O r>
(- 0 0C
PQ 4- 4
4 4J *0
o', 0-i 0
0 0 0 Q-I
ua Q- I -1
a) C *) 4
P 414 4
0 F- 0f
4-' 0 C:
0 0 0
0 0 0
) > d
4-4 H) m
0 0 0
S0 C 0
Cd (D U)
4- +4- 4+
0) 0) 0)
o o4 04
a) a) a)
-c-i 4- .-
0u 0 0O
a) ) a
o ) .
5-C 54 5-
a) a) a
-Ic +. +
r -4 -q
Q) Q) C) C) > >1 >1 >
4J 4-) 4-i -3 a) ) C)
r-l rl r4l -4 l 4 4 4 4
0 0 0 0 (1) (U ) )
r, mi I. r-c ,,D () () m
rY 10 CN M^ Ln m CN r-
(N O ,-I (N o m ( N Lr,
-KI -c ,N Co -K ---- P m
r O C Ci [- m0 0
Om o i o -i- c( N
La a a a tL) 1 -
b LA o ~-lo L 1
0 a0 r0l
3 tp a
>U C O) 4
*rl M0 r.
z 0 C 4 (d
XI *H 03*
z0 0rl (U
4) r-l 0
r )- i-
Hr U >1
H- 0 Q
0 ) 0) )
0 r a) 4 (U)
U0g H C on C U)
u 1 Q
- r- D C
0o l- k0 -1
0 N 0 0
co In r-I o
O C4 0 0
o o 00 d
Mo o o
,.I ,- -
r- r 1 -1 r-
1. 0 *-
0 O C
3 0 4-r
jH > 4
r-4 C) O
0 > a) f
4 1* 0
0 0 0
-4 04 04.
04 ) 0 0
(a> *-4 >
F 4 4,-
at En 0
(0 > > 0
(1- 0 0
4l u 0
2 4- O4-
0U r-I (
P0 M 4-4
44J U 0 4J
DECAY OF COLOSTRAL ANTIBODIES TO
BOVINE LEUKEMIA VIRUS
Before postnatal transmission of BLV can be studied,
criteria for serologic detection of BLV infection must be
established. This is straightforward for an animal which
seroconverts after a period of seronegativity. However,
during the period when colostral antibodies are detectable,
it is difficult to discriminate between infected and non-
infected calves strictly by the presence or absence of anti-
bodies. In order to more accurately define time or place of
infection, it would be helpful to detect infected calves
during the period that colostral antibodies are detectable.
This chapter will describe the decay of colostral BLV
antibodies in noninfected and infected calves and estimate
normal limits of decay in noninfected calves. It is shown
how these normal limits can be used to detect BLV-infected
calves during the first six months of life.
Materials and Methods
Titers were analyzed from calves which had colostral
antibodies on at least three observations, followed by no
serologic reaction for at least two months after the last
colostral antibody titer. These calves were assumed not to
be infected with BLV during the period when colostral anti-
bodies were detectable because, had the virus been present,
an active antibody response probably would have been ob-
served within two months after the last detectable colostral
titer (Mammerickx et al., 1980; Van Der Maaten and Miller,
1978b, 1978c). Furthermore, apparently to date, there is
no evidence that the presence of BLV appreciably alters the
rate of decay of colostral antibodies.
Titers also were analyzed from 14 calves which either
had precolostral antibodies that persisted throughout the
study or had antibodies persisting through eight months of
follow-up. These calves were assumed to have been infected
during at least part of the colostral-antibody period.
Only calves with titers which remained the same or declined
from previous levels were included in regression analysis.
All titers from infected calves were included in estimates
of ages at which BLV infection could first be serologically
Titers were analyzed using an estimated weighted re-
gression procedure (Swamy, 1971). All computations were
performed using SAS. A linear regression predicting the
logl0 of the inverse of BLV antibody titer at various ages
was determined by pooling intercepts and slopes from
individual calf least squares regressions. Each vector of
regression coefficients was weighted by the inverse of its
estimated covariance matrix. The mathematical model was
Y = a + bX + u
where Y (dependent variable) was logl0 of the inverse of
the end-point titer, a was the pooled Y intercept, b was
the pooled rate of decay of logl0 inverse titer, X (inde-
pendent variable) was age in days when the titer was ob-
served, and u was random error. Prediction bands (normal
limits) were calculated for the 90%, 95%, and 99% confidence
levels. A zero titer level was set at log10 (0.5) = -0.3.
Half-life of antibodies was estimated by
n log (10 ) a.
where a. and b. were the Y intercept and slope of the es-
timated regression line for the i calf and n = 130 calves.
A randomly selected set of 61 calves was used to ob-
tain preliminary estimates of a and b and normal limits.
The validity of these estimated normal limits of BLV anti-
body decay was tested using the remaining 69 calves as a
validation group. Since approximately 91%, 94%, and 98% of
the observations in the validation group fell within the
90%, 95%, and 99% normal limits established by the first
set of calves, the two sets were combined and a final pre-
diction model was derived based on all 130 calves.
Regression analysis was repeated for 14 infected
calves which had declining titers for at least three ob-
servations beginning at two days of age. Tests of hypothe-
ses of equal intercepts and slopes of colostral decay were
made using large approximate t statistics (Swamy, 1971,
For the 130 noninfected calves, the number of positive
antibody titers per calf ranged from three to eight. One
of these calves had an early titer as high as 1:256. Dura-
tion of colostral-BLV antibody titers ranged from 51 to 187
days with 50% of the calves serologically negative by 95
days of age (Fig. V-l).
The regression equation for all 130 calves was Y = 1.29
- 0.012X. The mean half-life of BLV antibodies was esti-
mated to be 27.1 days 1.2 days. The prediction line
crossed the zero titer level at 136 days of age, while the
90%, 95%, and 99% upper prediction bands crossed at 168,
178, and 200 days of age, respectively (Fig. V-2).
Serologic diagnosis of BLV-infected calves was made
between 2 and 180 days of age using the 90% limit and be-
tween 8 and 206 days of age using the 99% limit (Table V-l).
The regression line of BLV colostral antibody decay
for 14 infected calves was Y = 1.43 0.012X (Fig. V-3).
No differences were found between noninfected calves and
infected calves for slopes (p= 0.45) or intercepts (p=0.43).
Examination of individual calf regressions revealed
little deviation from linearity when titers were expressed
on the logl0 scale. The decision to set the level of a
negative titer at log10(0.5) was based on the two-fold
dilution scheme. Extrapolation below a titer of 1:1 can be
made with confidence because there is no reason to suspect
appreciable deviation from exponential decay below the
level detectable by AGID.
Some assumptions for estimated weighted regression
analysis (Swamy, 1971) could not be perfectly met in this
application. The distribution of log10 of inverse titers,
for example, cannot be assumed to be normal, particularly
as calves age, because titer values truncate at zero. Also,
the method of titer measurement results in a discrete de-
pendent variable which would not have an exact normal dis-
tribution. It is likely that with a sample of 130 calves
violations of these assumptions are not critical. Never-
theless, 69 calves were assigned randomly to a validation
set to test validity of the 90%, 95%, and 99% normal limits.
Since results showed little deviation from expectations, it
can be assumed that these limits were valid. Estimated
weighted least squares analysis was performed instead of
ordinary least squares regression because the assumption of
independence of deviations from overall regression does not
hold when repeated measures are made on each animal (Swamy,
It is well known that IgG is the major immune globulin
acquired from colostrum by the neonatal calf (Pierce and
Feinstein, 1965; Porter, 1972). These colostrum-derived
immunoglobulins originate from the maternal circulation
where they transfer to colostrum several weeks prior to
parturition (Brandon et al., 1971; Logan et al., 1973;
Smith, 1971; Sullivan et al., 1969). Since AGID using BLV
gp-51 antigen measures IgG1 exclusively (Matthaeus et al.,
1978), decay of colostral BLV antibodies is essentially the
decay of IgG The half-life of IgG1 in the bovine may vary
from 18 to 22.7 days (Brar et al., 1978; Dixon et al., 1952;
Logan et al., 1973; MacDougal and Mulligan, 1969; Smith,
1971). The longer half-life of 27.1 1.2 days found in
this study may have several explanations. Use of larger
and older animals in at least one study may account for a
shorter half-life, since younger and smaller animals may
metabolize IgG1 at a slower rate (Dixon et al., 1952).
Furthermore, the rate of decay of IgG1 may be dependent on
the incidence of diarrhea in calves studied (MacDougal and
The 187-day duration of colostral antibody titers to
BLV found in this study was similar to that for infectious
bovine rhinotracheitis (Brar et al., 1978) and bovine viral
diarrhea (Brar et al., 1978; Kahrs et al., 1966; Kendrick
and Franti, 1974). In another study of 21 calves, colostral
antibodies to BLV were detected for only 154 days (Fischer
and Keyserlingk-Eberius (1980). Since AGID has not been
standardized, a discrepancy of this magnitude is not sur-
prising. A lower rate of diarrhea or consumption of a
higher quality or quantity of colostrum also could explain
the longer duration of antibodies found in our study.
The age at which infected calves could be detected was
variable within a given level of normal limits. This is not
unexpected because the critical level at which viral repli-
cation is initiated also is variable and dependent on both
neutralizing antibody and virus concentration (Driscoll
et al., 1977; Straub, 1978b). At the 90% prediction level,
infected calves were identified as early as 2 days and as
late as 180 days of age. At the 99% level, calves were
identified later, from 8 to 206 days of age, but fewer non-
infected calves were misidentified. The sensitivity of
this method varies inversely, whereas specificity varies
directly, with the level of precision (prediction).
Serologic detection of BLV-infected calves during the
period of colostral antibody decay could be used in eradi-
cation or control programs where it would be feasible to
serially dilute serum from calves less than six months of
age. Because typical levels of passive antibodies vary
from region to region and herd to herd (Naylor, 1979), a
decay curve might be defined for noninfected calves which
is characteristic of a herd or region. Since at least three
positive observations are required per animal to estimate
error variance, calves could be bled at weekly intervals
for the first month and at monthly intervals thereafter.
This technique of defining upper limits of colostral
antibodies also could be used in planning strategies for
herd immunization (Kahrs et al., 1966; Kendrick and Franti,
1974). Calves would be vaccinated at the earliest age that
passive antibodies would not interfere with immunization.
Other factors, such as herd disease prevalence, antigen dose
and type, and general health conditions, also should be
considered (Brar et al., 1978; Kahrs et al., 1966; Kendrick
and Franti, 1974; Uhr and Bauman, 1961).
Colostral antibodies decayed at the same rate from the
same level in infected calves as in noninfected calves.
This validated the assumption that presence of BLV did not
alter BLV colostral antibody decay. Use of calves with no
evidence of BLV infection for at least two months beyond
the last colostral titer, therefore, should provide valid
estimates of BLV colostral antibody decay.
Table V-l. Ages at which bovine leukemia virus (BLV) anti-
body titers from BLV-infected calves were above
normal limits of colostral titers of noninfected
Age (days) at which logl0 1/titer* was first
above the normal limits of
calf number 90% 95% 99%
J166t 2 2 178
J171 153 153 153
J183$ 151 151 151
Jl94Bt 169 169 169
J200Bt 38 73 206
H875t 180 180 180
H888 154 154 154
H913$ 155 155 155
H938 112 112 167
H949 2 2 8
H955t 160 160 160
H972 137 137 163
H973 67 67 102
H1002Bt 31 31 31
H1038Bt 27 27 144
H1079B 167 167 167
MEAN 106.56 108.75 143.00
SEM 16.54 16.07 13.22
Antibodies measured by agar-gel immunodiffusion using
tCalves with precolostral antibodies to BLV.
tCalves with no detectable precolostral antibodies to BLV.
Calves which were not sampled precolostrally.
- - -
20 40 60 10 1 l 10 1? 2)t
CALF AGE (DAYS)
Fig. V-1. Percentage of 130 calves with colostral anti-
bodies to bovine leukemia virus (BLV) as a
function of age.
> 0 0.6
- 95% LIMITS
...... 90% LIMITS
----- EXTRAPOLATED LINE
AAA AA A A A
0AooAoA '0 00
CALF AGE (DAYS)
Prediction line of the decay of colostral anti-
bodies to bovine leukemia virus (BLV) in 130
calves with no evidence of infection with BLV.
Upper normal limits of the prediction line are
indicated for the 90%, 95%, and 99% confidence
levels; o=values of logl0 1/titer falling
above the 90% normal limit from calves not in-
fected with BLV; A=values of logl0 1/titer from
calves infected with BLV in utero or sometime
prior to six months of age.
- PREDICTION LINE
------ EXTRAPOLATED LINE
0 a 0
* ^s* *
*** ** *** *
, 00 ~
20 40 6 0 100 120 1
20 40 60 80 100 10 140
160 1 0 200 210
CALF AGE (DAYS)
Prediction line of the decay of colostral anti-
bodies to bovine leukemia virus (BLV) in 14
calves infected with BLV in utero or sometime
prior to six months of age; =values of log10
AGE-SPECIFIC RATES OF DETECTION OF
BOVINE LEUKEMIA VIRUS INFECTION
Several factors have been examined individually for
association with prevalence rates of BLV infection in previ-
ous studies. Rates have been reported to be unrelated to
BLV-status of the dam (Hofirek, 1980; Olson et al., 1978;
Valikhov, 1978), consumption of colostrum from BLV-infected
cows (Ferrer et al., 1976; Ferrer and Piper, 1978; Piper
et al., 1975, 1979), or breed (Burridge et al., 1979).
Other studies have suggested that an association exists
between prevalence rates of BLV infection and breed
(Burridge et al., 1981), BLV status of the dam (Baumgartener
et al., 1978), type of colostrum consumed (Seger and Morgan,
1977), and age (Burridge et al., 1979, 1981; Chander et al.,
1978; Evermann et al., 1980; Ferrer et al., 1976; Hofirek,
1980; Huber et al., 1981;Mammerickx et al., 1978a, 1978b;
Marin et al., 1978; Olson et al., 1973; Parfanovich et al.,
1978; Piper et al., 1979). Because age is associated with
BLV prevalence rates, it is important to examine for the
age-specific effects of factors on rates of infection.
The objective of this chapter is to describe age-
specific rates of detection of BLV infection and to test
for differences between rates for cattle identified by
various maternal and management features.
Materials and Methods
Cattle Studied and Criteria for Detection
of BLV Infection
Cattle studied were those described in CHAPTER III.
Calves known to be infected in utero (CHAPTER IV) were ex-
cluded from examination of postnatal infection. Detection
of BLV infection was made when BLV-gp antibodies first ap-
peared following a period of seronegativity. In calves
showing evidence of colostral BLV antibodies, e.g., a declin-
ing titer during the first six months of life, the time at
detection of BLV infection was determined from the model of
colostral antibody decay (CHAPTER V). A calf was detected
as infected when titers first began to increase or when a
titer fell outside the 95% normal limits of colostral anti-
body decay, whichever occurred first. Because the precise
moment of infection could not be measured, the date of de-
tection was used to approximate this time, assuming a uni-
form lag period from infection to detection in all cattle.
A BLV-infected dam was defined as one which had anti-
bodies to BLV during gestation or within one month follow-
ing parturition. A calf was judged to have consumed colostrum
from a BLV-infected cow if the calf had detectable anti-
bodies to BLV in the first postcolostral sample and was
precolostrally negative or if it had postcolostral antibody
titers that declined with age.
Data were analyzed using survival regression tech-
niques (Cox, 1972; Taulbee, 1979) available in the PHGLM
program of SAS. Life-table calculations and graphics were
performed using the statistical program P1L of BMDP. Fail-
ure time was defined as the age in days when an animal was
detected as infected with BLV. Censored time was the age
when an animal died, left the herd, calved, or when follow-
up was terminated on September 30, 1981, whichever occurred
The number of cattle at risk of being detected as in-
fected in an age interval ti, defined for graphical purposes
as 30 days, was
r. = n.
1 1 2
where n. was the number of animals entering the interval and
c. the number censored in the interval. Conditional prob-
ability of detection in an interval (given that detection
did not occur in the previous interval) was
where d. was the number detected as infected in the interval.
It was assumed that risk of detection remained constant
during an interval. The conditional probability of an
animal not being detected (surviving) in the ith interval
pi = 1 q
The cumulative proportion remaining undetected (also known
as cumulative survival or survival rate) to the beginning
of the i interval was
Pi = Pi-Pi-
where P =1. Hazard rates (also called failure rates or
force of morbidity) were estimated at interval midpoints
and were defined as the quantity
Xi = 2qi/hi(l+pi)
where h. was the width of the i interval. This rate was
a linear interpolation estimate based on a life table rep-
resentation of a survival curve.
Standard errors of cumulative survival functions and
hazard rates were calculated as described in the P1L pro-
gram of BMDP for all animals born between July 1979 and
June 30, 1981.
Relationships between hazard rates and other variables
were assessed using the model of the hazard at time t, as
previously described (Cox, 1972) because it adjusts for
censoring in the data. This model is
A(t,zi) = exp(z.i)A(t) ,
where 8 is a vector of unknown parameters and z is the vari-
able under study. The variable z can take on indicator
values (e.g., 1 =positive dam and 0 =negative dam) or
continuous values as in dam age. Implicit in this model
is the assumption that hazard rates for individuals with
different values of the covariate are in constant ratio
over time, regardless of the underlying hazard rate
(Taulbee, 1979). An obvious violation of the proportional
hazards assumption occurs when survival curves cross. This
often is the case with actual data, and it is possible to
examine changes in hazard rates with age using an alterna-
tive function described by Taulbee in which a new variable,
zt, is defined. This allows the ratio of hazard functions
to differ and permits examination for age-by-factor inter-
actions. Conclusions based on this test, however, must be
extremely tentative, since the differences detected as
significant are unknown.
Correlation coefficients were calculated using the
CORR program of SAS to describe relationships between the
various factors investigated. To examine effects of factor
interactions on survival, correlation coefficients were
computed from the variance-covariance matrix of parameter
estimates when all factors were included in the survival
model. High negative correlations between regression pa-
rameters for two factors were interpreted as a reduction in
the effectiveness of one factor in explaining survival when
the role of the other factor on survival became more
important. High positive correlations identified factors
which were jointly effective in explaining survival.
Factors examined were dam age (months), dam parity
(0 =primiparous, 1 multiparouss), dam BLV status when the
animal under study was born (1 =positive, 0=negative),
type of colostrum consumed (1 = from a BLV-infected cow,
0=not from a BLV-infected cow), and breed (1= Hostein,
A total of 473 live calves entered the survival analy-
sis at birth. Of these, nine were detected as infected in
utero (CHAPTER IV) and 54 infected postnatally. Two ani-
mals were followed through 780 days of age (Table VI-1).
Estimates of prevalence rates of infection increased from
2.3% at birth to 63% at 27 months of age (Fig. VI-1).
Survival rates (Fig. VI-1), hazard rates (Fig. VI-2), and
proportion of animals detected as infected for various age
groups (Fig. VI-3) demonstrated four possible age-related
stages of detection. The first stage, prenatal infection,
has been characterized previously (CHAPTER IV). A second
stage appeared as increased hazard rates during the first
six months of life. This was followed by a 10-month period
of sporadic detection. The fourth stage, beginning at 16
months of age, was characterized by sharply increasing
hazard rates of detection through 27 months of age.
Results of statistical analyses using Cox's procedure
are presented in Table VI-2. No differences were found in
hazard rates when variables were examined separately without
the age covariate. However, significant age interactions
were suggested for all factors studied after inclusion of zt
and use of the test by Taulbee (Table VI-2). Graphically,
factor-by-age interactions were indicated for calves con-
suming BLV-positive colostrum (Fig. VI-4), for dam status
(Fig. VI-5), for breed (Fig. VI-6) and for dam parity
(Fig. VI-7). Lower rates of detection of BLV infection
were observed in young calves from infected dams and in
calves consuming colostrum from infected cows. Higher rates
of detection were observed in older Jerseys than in Holsteins
of the same age. Calves born to primiparous dams appeared
to have lower rates of detection through 18 months of age,
after which they began to experience higher rates of detec-
tion until 27 months of age when rates were similar.
Significantly high factor correlations were found be-
tween type of colostrum consumed and dam status (p=0.0001)
and between dam age and parity (p =0.0001), and low, but
significant, correlations were found between dam parity and
dam status (p =0.0001), between colostrum and parity (p=
0.001), and between colostrum and dam age (p =0.001) (Table
VI-3). High, negative correlation coefficients of estimated
parameters of survival were found only for dam age and
parity (r=-0.97) and for colostrum and dam status (r=
-0.86) (Table VI-4).
Throughout this study increasing or persisting antibody
titers to BLV, as measured by AGID-gp, were interpreted as
evidence for BLV infection. Because this method indirectly
tests for presence of virus by an immune response, some as-
sumptions regarding sensitivity and specificity of the
diagnostic criteria should be made.
The BLV-gp antigen appears to be unique, since no cross
reactivity with proteins of other oncoviruses or retro-
viruses has been demonstrated (Ferrer, 1972; Ferrer et al.,
1975; Kaaden et al., 1977; McDonald and Ferrer, 1976;
McDonald et al., 1976). A positive serologic response to
BLV-gp, therefore, can be considered specific for BLV.
Comparisons of several serologic tests have demonstrated
that AGID-gp is a relatively sensitive serologic test (Burny
et al., 1980). These results are corroborated by experi-
mental studies in which a serologic response following BLV
inoculation was consistent (Miller et al., 1972; Schmidt
et al., 1976; Van Der Maaten and Miller, 1978b; Van Der
Maaten et al., 1981a) and persistent (Miller and Van Der
Maaten, 1976b). Similarly, BLV has been readily detected
from seropositive but not seronegative animals following
natural infection (Ferrer et al., 1976, 1977a, 1977b;
Olson et al., 1973; Piper et al., 1979).
There is no evidence for recovery from BLV infection.
On the contrary, seropositive cattle older than six months
appear to maintain detectable antibodies throughout life
(Chander et al., 1978; Kaaden et al., 1978; Tabel et al.,
1976). In a study to examine vaccine efficacy, inactivated
BLV produced only a transient serologic response, and virus
was not isolated (Miller and Van Der Maaten, 1978b). In
that study two doses of killed virus were required to induce
a serologic response. This suggests that persistent anti-
body titers only result from active, persistent infection
and that any transient serologic response resulting from
exposure to a killed virus is rare.
Until recently, it could not be stated that absence of
either virus or antibody implied absence of specific BLV
sequences in host cell genomes. Studies now suggest that
such sequences are not to be found in cattle without BLV or
specific BLV antibodies (Callahan et al., 1976; Deschamps
et al., 1981; Kettmann et al., 1976).
A final demonstration of the sensitive nature of AGID
using gp-51 lies in results of BLV-eradication programs
based upon mass screening with this test. Results of these
campaigns indicated that the virus was efficiently removed
from aherd by test-and-slaughter methods (Mammerickx et al.,
1978a; Schmidt et al., 1978, Straub,1978b). If AGID were
not an efficient discriminator of infected and noninfected
cattle, it is unlikely that eradication efforts would have
met with such rapid success. Use of AGID-gp to detect BLV
infection, therefore, is justified, particularly in a design
in which animals can be followed for several months and
persisting titers can be observed.
Survival analysis and life-table methods are well es-
tablished techniques in epidemiology and actuarial science
(Gross and Clark, 1975; Lilienfield, 1976). Recently these
methods have seen increased application in veterinary
epidemiology (Cobo-Abreu et al., 1979; Hird et al., 1975;
Schwabe et al., 1977). Survival analyses are especially
appealing in prospective studies where animals lost to
follow-up can be considered in computations up to the time
of censoring, assuming failure rates similar in censored
and uncensored animals.
Data presented in this study are particularly suitable
for examination by survival methods for two reasons. First-
ly, recovery from BLV infection is unknown and animals can
be defined as truly failed once infection is detected.
This is analogous to death in traditional applications.
Secondly, since these methods incorporate the dimension of
time, either by comparing distributions over time or by
using time as a dependent variable, they are appropriate
in studies of BLV infection because prevalence rates are
known to be associated with age (Burridge et al., 1979,
1981; Chander et al., 1978; Evermann et al., 1980; Ferrer
et al., 1976; Mammerickx et al., 1978a, 1978b; Marin et
al., 1978; Olson et al., 1973; Piper et al., 1979).
Cox's model was used to test estimated parameters of
a survival function where the underlying assumption was
that hazard rates were proportional (Cox, 1972; Taulbee,
1979). The model tends to force nonparallel curves to a
parallel model. Thus, interaction, characterized by cross-
ing or bellying curves, could not be examined. Inclusion
of the zt variable removes dependence on proportionality and
allows for tests of interactions with age (Taulbee, 1979).
Estimated age-specific prevalence rates were calculated
as one minus the proportion remaining negative for BLV in-
fection. The prevalence rate estimate for 18-month-old
cattle of 25% was lower than that of 45% described from a
cubic polynomial curve fitted to data collected from the
same herd for the years of 1975 through 1977 (Burridge
et al., 1979). Transmission rates of BLV infection, there-
fore, appeared to have declined over the past four to five
years, at least in animals up to 18 months of age. This
may suggest variations of prevalence rates in yearly birth
cohorts, as found in other prospective studies (Huber et al.,
1981; Wilesmith et al., 1980). In those studies, it was
observed that cattle born in a given year experienced a
different life-time prevalence rate than did animals born
in some other year. Furthermore, the prevalence rate
within a cohort did not change appreciably beyond two
years of age. A similar situation may have occurred in
this herd due to management of Holstein heifers. Prior to
1977, bred Holstein heifers were kept in the dry herd until
shortly before calving. In the past few years, these
heifers have been managed separately from the dry herd,
whenever pasture space permitted, in order to provide them
with a better plane of nutrition. In this study Holstein
heifers were exposed to the dry herd for a relatively short
time, January through April 1981, which could have limited
BLV transmission and resulted in the lower prevalence rates.
Inspection of the overall life table revealed heavy
censoring during the first two to three months. This was
due to sales of bull calves and to neonatal mortality.
Early postnatal detection of BLV infection, constituting
the second detection phase, could have been represented by
calves infected in utero and not sampled precolostrally or
infected too late in gestation for antibodies to appear by
birth, as previously suggested (Van Der Maaten et al.,
1981b). Evidence that postnatal infection occurred during
this period was demonstrated by detection of infection in
calves born to noninfected dams.
The length of this period of detection would not cor-
respond to the length of an underlying infection period
because of the seroconversion lag time (CHAPTER VIII) and
possible repression of viral expression by colostral anti-
bodies. Inhibition of release of BLV by antibodies has
been demonstrated in vitro (Driscoll et al., 1977) and also
suggested in studies of colostral antibodies (Van Der Maaten
et al., 1981a). Antigenic modulation or inhibition of P
proteins (Sissons and Oldstone, 1980) may explain such a
phenomenon. The process of modulation is believed to
involve a binding of viral glycoprotein on the cell membrane
by circulating antibodies. Glycoprotein surface markers
become masked from recognition by K cells or other cells
invoking an antibody dependent cytotoxic response. A
failure to recognize virus-bearing lymphocytes by the
immune system renders them resistant to immune damage. If
antibodies are removed, for instance through natural at-
trition of gammaglobulins, the lymphocyte would regain the
ability to shed virus. Inhibition of P proteins may also
be involved in repressing virus expression by inhibiting
A further delay in detection, beyond the two- to three-
month seroconversion period and one-month sampling interval
could have resulted from early sampling-design flaws. As
much as a two-month delay could have occurred as a result
of the three-month sampling interval used at the beginning
of the study. It is likely, therefore, that infection
represented by this phase of detection occurred in the
first two to three months of life or even prenatally.
Several factors may have contributed to postnatal
transmission in the first two to three months. Calves may
have become infected at parturition when a calf born to an
infected dam ingested maternal blood from uterine or vaginal
tears (Van Der Maaten et al., 1981b). Rates of detection
in calves from infected dams, at least during the first
three months, however, were not different from rates in
calves from noninfected dams. Such a similarity of rates
may be an artifact due to delayed expression of the virus
by colostral antibodies. Consumption of colostrum with
BLV antibodies, therefore, may complicate control efforts
by masking early infection.
Type of calf management also may have influenced BLV
transmission during the first few months of life. Most
calves were placed inside the calf barn where they could
have been exposed to infected calves in adjacent pens.
Transmission may have resulted from ingestion or inhalation
(Van Der Maaten and Miller, 1978c) of saliva (Ressang et al.,
1980), urine (Gupta and Ferrer, 1980), blood (Mammerickx
et al., 1980; Van Der Maaten et al., 1981a), or lymphocytes
(Miller and Van Der Maaten, 1978b; Van Der Maaten and
Miller, 1978b, 1978c, 1981a). Animals were also tattooed
while in the barn, introducing the opportunity for iatro-
genic transmission. Other calves occupied reasonably iso-
lated pens outdoors, and they may not have experienced the
same risks of infection as indoor calves. Risks for animals
in various locations are examined further in CHAPTER VIII.
A third relatively quiescent phase of detection of BLV
infection between 6 and 15 months of age coincided with
movement of animals from densely occupied paddocks to
fields and pastures (APPENDIX D, Fig. 7). It also was noted
that vaccinations for other infectious diseases were given
either shortly before or during this period. Iatrogenic
transmission of BLV, therefore, does not appear to have
been of major importance in transmission because rates of
detection actually declined during this period. A more
detailed examination of detection rates and vaccination pro-
cedures is presented in CHAPTER VII.
A sharp increase in detection rates typified the fourth
and last observed phase which began after 15 months of age.
The only management interventions preceding the increased
rates were insemination and movement of bred heifers to the
dry herd. Infection as a result of close physical contact
has been suggested as the major mode of BLV transmission
(Dechambre et al., 1968; Maas-Inderwiesen et al., 1978;
Straub, 1971, 1978a; Miller and Van Der Maaten, 1978a;
Wilesmith et al., 1980). Other reports mentioned similar
observations or seroconversions following exposure to in-
fected animals (Ferrer et al., 1976; Piper et al., 1979).
However, in our study not all bred heifers entered the dry
herd, permitting a comparison of risks of two types of
heifer management in CHAPTER VIII.
It is not likely that transmission resulted from in-
semination, since the exogenous nature of BLV excludes
transmission via gametes. This has been confirmed by
observational studies which did not find increased infec-
tion rates in animals sired by BLV-infected bulls (Baum-
gartener et al., 1978) and by experimental studies which
failed to transmit BLV with semen from BLV-infected, but
otherwise healthy, bulls (Miller and Van Der Maaten, 1979;
Ressang et al., 1980). Furthermore, the time from breeding
(13-16 months of age) to peak rates of detection (18-21
months of age) was longer than suggested by experimental or
natural transmission studies (Gentile and Rutili, 1978;
Mammerickx et al., 1980; Straub, 1978b; Van Der Maaten and
Miller, 1978a, 1978b). Unfortunately, an effect of in-
semination on transmission could not be examined further
because BLV status of sires was unknown and analysis would
have been confounded by the presence of only bred heifers
in the dry herd.
In the examination for associations between hazard
rates and dam age, dam parity, dam status, breed, and colos-
trum type using Cox's model, no significant main effects
were observed. Results of Taulbee's alternative procedure,
however, suggested significant departures from proportional
hazards. The sizes of the differences detected by this al-
ternative method are not known, and they may be too small
to be of practical importance in the control of BLV in-
fection. Conclusions regarding the practical significance
of interactions, therefore, should be very tentative, but
discussion can be based on inspection of survival curves.
Graphical examination is meaningful because crossing of
survival curves cannot occur without crossing of hazard
curves, thus allowing for visual inspection of factor-by-
Interactions between age and dam status and between
age and colostrum have already been discussed and indicated
that calves which consume colostrum from an infected cow or
which were born to a seropositive dam were protected from
detection but not necessarily infection, at an early age.
These two factors would be expected to have similar curves
because of the high negative correlation of their estimated
survival parameters. Although several factors were sig-
nificantly correlated with each other, only dam parity and
dam age, and dam status and colostrum were correlated when
hazard rates of detection were considered.
An interaction of breed by age was suggested when old
Jersey heifers were observed to have higher rates of detec-
tion than old Holstein heifers, but at young ages curves were
similar. It has been reported that Jerseys have low levels
of circulating gammaglobulins (Logan et al., 1981), but
this would not necessarily explain increased susceptibility
only in old Jersey heifers. A more reasonable explanation
for lower survival rates in old Jersey heifers is that they
remained in the dry herd until shortly before calving.
Holstein heifers, on the other hand, were removed from the
dry herd on about May 1, 1981, and Holsteins bred 45 days
prior to that time were never in the dry herd. The breed-
by-age interaction, therefore, was due most likely to
longer exposure of Jerseys to older, infected cattle in
the dry herd.
There has been no mention made in the literature of
dam parity or dam age effects on subsequent BLV infection
of progeny. Crossing of survival curves of animals from
primiparous and multiparous dams observed in this study
suggests an age-by-parity or age-by-dam age interaction.
Whether the crossing of these curves represents expected
variation, especially with small numbers at older ages, or
a real effect cannot be stated. No other explanation can
ODn OOOOOO 1-
L(N min m
0 .o o0
LOn -I Ln r-
" i-i Ln %10 .-i 1, 'o 0 0 -i 0 (N
Nr Co r-I
Lfn 00 r-4 0 O)
( r- r- N
CN ,-I r- r-
1-4 -4 1-4 r-i r-4
N Ln co
-I0 O0 N
co Lfn 1- (N ^r rN
0 on -I cm (N
Om, CO 00
-N -4 -4
*i N 0 m4
m m 1 M r-
H4 HH 04 H
00000; ( C
v i LnA nC4 0
n N 1-1 r- m 0
m o NT 0 ; v M Ln
o r-o0 -o0 0 o
mO f Or
000000 0000 00000000
m 0r 00 oo
'. C4 0D C4' N L 0C4 'D
0N co LAn c LoA OI 4rn oo
Ln 04 tL 04 LA k.0 N- '. 0
O mo Loin z1 m
O 04 H O A m L cN 0C 0 -H 0
C H-- 0N
-lH l- r-
H- 0N M M LAn LA 00 CM
Nr Ln m 0) CN 0 m m H (v)
0' 2 co '. w0 LA O
01 0 00 0 CO
U U) LA C> \ 0 n
r r- r- 113 in L o
"1-4 0 .
> -I o 0 0 0 0
_ II II II II II
S 0 o 1 1 LA o
i4 .- 0 L n 0
O O o O O O4
0 n O o o o o
& > O O O o O
0 0 0 0 0 0
s C) C 0
o k o
U) C 0 0 0
4-) >1 0 0 0 0 0
a 0 0 0 0 0
*-H 0 '- I r-I r-l ,-1
S0 0 CN 0 0 0
0I 0C C
4- O 2
- (U0 *) U)
S(q < ( C O )
5- E -4 U )
M (i (d- 0 5- (
E- > Q m Q U m
Table VI-4. Correlation coefficients of estimated parame-
ters of hazard rate functions for the detection
of bovine leukemia virus infection.
Variable Dam status Dam age Colostrum Breed
Dam parity -0.002 -0.973 0.027 0.087
Dam status -0.027 -0.857 -0.021
Dam age -0.036 -0.052
.--- CUMULATIVE PREVALENCE
Z 60- /
I I I I I I 1 1 1 1 I I I 1 1 1 I 1 I i
3 6 9 12 15 18 21 24 27
Fig. VI-1. Estimated age-specific cumulative survival
(%) and prevalence rate (%) of detection of
bovine leukemia virus infection.
3 6 9 12 15 18 21 24
Fig. VI-2. Age-specific hazard rates of detection of
bovine leukemia virus (BLV) infection.
9 12 15 18
AGE (MONTHS) AT DETECTION
Age at detection of bovine leukemia virus (BLV)
infection for 63 BLV-infected animals.
1.0 ---- CALVES CONSUMING
BLV- POSITIVE COLOSTRUM
.9 ------------CALVES CONSUMING
S. -------- BLV-NEGATIVE COLOSTRUM
Fig. VI-4. Cumulative proportion of calves consuming colos--
antibodies remaining undetected for BLV infec-
100 200 300 400 500 600 700 800
Fig. VI-4. Cumulative proportion of calves consuming colos-
trum with bovine leukemia virus (BLV) antibodies
and of calves not consuming colostrum with BLV
antibodies remaining undetected for BLV infec-
CALVES BORN FROM
------CALVES BORN FROM
Cumulative proportion of animals from bovine
leukemia virus (BLV)-infected dams and from
noninfected dams remaining undetected for
Cumulative proportion of Holsteins and Jerseys
remaining undetected for bovine leukemia virus
9- ----BORN TO
o MULTIPAROUS DAMS
I-1 I I I I I I I I I I I-
100 200 300 400 500 600 700 800
Fig. VI-7. Cumulative proportion of animals born to primi-
parous cows and of animals born to multiparous
cows remaining undetected for bovine leukemia
virus (BLV) infection.
SEASONAL PATTERNS OF RATES OF BOVINE LEUKEMIA
A fundamental consideration in the epidemiology of in-
fectious diseases is the examination for temporal patterns
of occurrence of a disease. Seasonal or secular trends in
incidence rates may suggest modes of transmission, causal
associations, and/or etiologic agents. If, for example, a
hypothesized factor in disease transmission demonstrates
seasonal variability, and a statistically significant season-
al cycle of disease exists in phase with the hypothesized
factor, then supporting evidence favoring involvement of
the factor in transmission has been demonstrated. On the
other hand, if no seasonal occurrence of disease can be
demonstrated, or it is not in phase with that of the factor,
then the hypothesis of factor association should be rejected.
Interest in possible vector-borne transmission of BLV
has arisen from evidence for arthropod involvement in the
spread of another retrovirus, equine infectious anemia virus
(Fischer etal., 1973; Hawkins et al., 1973a, 1973b). Re-
ports based on experimental and observational data have
suggested the involvement of arthropod vectors in BLV trans-
mission (Bech-Nielsen et al., 1978; Ohshima et al., 1981).
Another study, however, revealed higher rates of infection
during winter months, coinciding with crowded, indoor hous-
ing conditions rather than exposure to large numbers of
flies (Wilesmith et al., 1980).
In this chapter, the hypotheses tested are as follows:
(1) age-specific rates of detection of BLV infection do
not differ with respect to month of birth; (2) rates of
detection of BLV infection show no significant seasonal
pattern associated with the seasonal frequency of potential
arthropod vectors; and (3) no other seasonal pattern of
rates of detection exists.
Materials and Methods
Month of Birth
Cox's hazard model described in CHAPTER VI was used to
examine for an association between month of birth and sub-
sequent BLV infection. Month of birth was denoted by z in
the model. Detection of BLV infection was based on cri-
teria presented in CHAPTER VI.
Month of Detection of BLV Infection
Monthly incidence rates of detection of BLV infection
were computed to estimate seasonal patterns of BLV infec-
tion rates. Monthly incidence rates of detection were cal-
culated as the number of animals detected during the month
divided by the number of animals at risk of being detected
in that month. The criteria for an animal to be at risk for
any month of follow-up were (1) the animal must not have
been detected previously; (2) the animal must have been
present for at least half of a month; (3) an animal born in
the last half of a month was considered only in subsequent
months; and (4) an animal remaining undetected for more
than one year would be considered also for each repeated
month. For example, if an animal were present during
January 1980, and January 1981, it would be counted as
two animals-at-risk during the month of January. Those
animals detected as infected during a month also were
counted in the at-risk denominator.
Possible differences between the 12 monthly incidence
rates were examined using a Chi Square test with 11 degrees
of freedom. In order to determine if monthly incidence
,rates followed a simple harmonic trend, a test for goodness-
of-fit of the data to a simple harmonic curve was performed
using a Chi Square statistic, as previously described
(Walter and Elwood, 1975). The test for a simple harmonic
trend of incidence rates, however, could not be pursued
because a high Chi Square value (X = 17.5, 0.05 (11)
indicated that these rates were inappropriately described
by a simple harmonic curve.
An alternate, nonparametric approach to the seasonality
of events was used to examine for six-month trends in in-
cidence rates (Hewitt et al., 1971) Rates were computed
only for animals over 12 months of age. The tested
hypothesis stated that six-month cumulative ranks of inci-
dence rates from May through October, the fly season, were
not greater than those for the 11 other six-month periods.
A second hypothesis was tested for no differences in cumu-
lative ranks of incidence rates for any nonspecified six-
Month of Birth
No significant differences were found between hazard
rates of detection of BLV infection for animals born in
each of the 12 months (p= 0.24). The proportion of animals
infected was highest in animals born in August, September,
and October (Fig. VII-1).
Month of Detection of BLV Infection
No significant differences were found between the 12
monthly incidence rates (X(II) = 9.81, 0.4
The numbers of animals detected as BLV infected and at
risk of detection for each of the 27 months of observation
are presented in Table VII-1. Monthly incidence rates of
detection for animals of all ages ranged from 1.1% in
January and April to 3.0% in June (Fig. VII-2). For ani-
mals over 12 months of age, cumulative ranks of monthly
incidence rates between May and October were not signifi-
cantly greater than those for any other six-month period
(p= 0.20) (Table VII-2). Cumulative ranks of monthly in-
cidence rates were highest for the six-month period be-
ginning with March and were lowest for the period beginning
with September. The probability of cumulative ranks was
lowest for the period beginning with March, but this was
not significant without an a priori commitment (p= 0.13).
Although no statistical significance could be placed
on differences between birth-cohort hazard rates, calves
born between August and October constituted 65% of the in-
fected animals, even though only 42% of calves were born
during that period (APPENDIX C). This was not surprising
for several reasons. The density was highest during the
period in which these calves were in the calf barn
(CHAPTER III). In a crowded situation, transmission may
occur either by aerosol (Van Der Maaten and Miller, 1978c)
or by contact (Dechambre et al., 1968; Maas-Inderwiesen
et al., 1978; Miller and Van Der Maaten, 1978a; Straub,
1971, 1978a; Wilesmith et al., 1980). Also, since these
animals were the oldest in the study, they had a greater
chance of entering the dry herd where there may have been
an increased risk of infection due to contact with older,
BLV-infected cattle. Calves born during the winter and
spring were too young to be bred and enter the dry herd by
May 1981. The proportion of calves born in the spring and
subsequently becoming infected would be expected to be low
because few cows calved at that time (APPENDIX C).
Two assumptions were made in the analysis of monthly
incidence rates of detection of BLV infection. One was
that animals were considered independent within and among
months followed. This is reasonable because exposure to
a possible BLV-carrying insect and susceptibility to BLV
infection were not likely to be related to any previous
The other assumption in examination of monthly inci-
dence rates was that any pattern of detection would be
synchronized with that of infection and would consistently
follow infection by two to three months. This assumption
may not be valid for calves less than six months of age.
Infection may occur in utero or shortly before birth but
before acquisition of passive protection from colostral
antibodies. In such cases, detection of infection may not
be made for six months or more because virus repression may
be dependent on both antibody and virus concentration
(Driscoll et al., 1977; Straub, 1978a). It is believed,
however, that viral expression would appear within 12
months of infection (Van Der Maaten et al., 1981a).
Analysis of possible vector transmission also may have
been confounded by calfhood management. Since the calving
season occurred during the summer and early fall, animal
density in the calf barn was highest during late summer and
fall. Seroconversion of animals several months later could
be interpreted as infection resulting from close physical
contact rather than from arthropod vectors. Furthermore,
the risk of infection for calves in the calf barn may not
have been as high as for those over one year of age. Flies
found in the calf barn were not usually the biting type,
and sporadic insect control was practiced, which limited
exposure to potential vectors. Colostral antibodies also
may have provided some resistance to infection from small
doses of virus introduced by an insect (Van Der Maaten
et al., 1981a).
The situation with cattle more than 12 months of age
was quite different. Biting flies were present on these
animals throughout summer months, and no fly control was
practiced. Prevalence rates of infection also were higher
in these animals (CHAPTER VI), increasing the chance that
a fly would first feed on an infected animal. Cattle over
one year of age, therefore, were analyzed separately as they
constituted a more susceptible and exposed population to
possible vector-borne transmission and because infection
was more easily detected and analysis was not confounded
by management factors.
Results of the Chi Square test indicated that there
were no significant differences between monthly incidence
rates. Important seasonal trends, however, may be present
without apparent rate differences between months (Walter
and Elwood, 1975). The method proposed for analysis of
seasonal trends in incidence rates (Edwards, 1961; Walter
and Elwood, 1975) was particularly appealing because season-
al frequencies of potential arthropod vectors in temperate
and subtropical climates generally resemble simple harmonic
curves which peak during summer months (Beck, 1958; Blanton
and Wirth, 1979; Bohart and Washino, 1978; Jones and Anthony,
1964; Khalaf, 1969). Their analysis could not be pursued
because the incidence rates were inappropriately described
by a simple harmonic curve. The goodness-of-fit test in-
dicated that incidence rates did not resemble a simple har-
monic curve and no further tests for significance or maximum
amplitude of a curve should be performed. This analytic
method also could not be used to test rates in older animals
because the recommended sample size is not less than 50
(Walter and Elwood, 1975). Instead, a nonparametric method
appropriate for small samples was used to examine for ex-
cessive rates over six-month periods (Hewitt et al., 1971).
Difficulties with Hewitt's alternative procedure have
been its lack of power and inability to estimate parameters
of an harmonic curve (Walter and Elwood, 1975). In spite
of these restraints and the few infections detected, a
marked tendency for maximum risk of detection of infection
was revealed between March and August (p= 0.13), the six-
month period beginning with March. Had an a priori hypothe-
sis been specified for that particular yearly segment, there
would be reason to believe that an excessive rate of de-
tection had occurred during that period (p= 0.01). This
interval would coincide with infection resulting from
exposure to animals in the dry herd between January and May
1981. If vectors had played a major role in transmission
and assuming peak infection around July, then maximum
cumulative ranks of incidence rates of detection should
have been observed between May and October. This was not
the case, as indicated by acceptance of the hypothesis of
no increase in incidence rates for that six-month period.
Results of this study are consistent with those from
a study in which increased rates of seroconversion were ob-
served after cattle grazed together on summer pastures
(Onuma et al., 1980). In that study, arthropod-borne
transmission could not be examined because analysis would
be confounded by congregation of cattle only during the
vector season. In another study, two groups were compared
for rates of infection during winter and summer months
(Bech-Nielsen et al., 1978). Rates observed for summer
and winter groups, 4/7 and 1/7, respectively, could have
occurred by chance (Fisher's exact p=0.13). A study of
incidence rates of BLV seroconversion in a large dairy
found higher rates in winter and spring, implying trans-
mission associated with crowded housing (Wilesmith et al.,
1980). None of these studies examined monthly rates for a
complete seasonal cycle. Future studies of seasonal patterns
of disease should be based on at least two years of monthly
cohort data in an attempt to remove as much year-effect as