THE JOURNAL OF INFECTIOUS DISEASES • VOL. 131, NO.5· © 1975 by the University of Chicago. All rights reserved.
MAY 1975
Cervical Cytomegalovirus Excretion in Pregnant and Nonpregnant Women: Suppression in Early Gestation Sergio Stagno, David Reynolds, Alex Tsiantos, * David A. Fuccillo, Richard Smith, Mary Tiller, and Charles A. Alford, Jr.
Comparison of 659 pregnant and 202 nonpregnant women with similar demographic characteristics showed overall rates of cervical cytomegalovirus excretion that were identical (9.5% vs. 9.4%) and were surprisingly high, especially since 89% of the pregnant group possessed antibody to cytomegalovirus when admitted to the study. Prevalence of cytomegalovirus among gravidas was significantly lower during the first (1.6%) than during the third (11.3%) trimester. Thus, early pregnancy appeared to exert a suppressive effect on viral excretion that waned with advancing gestation. A similar but less significant occurrence was observed in the two groups with respect to viruria. Increasing age also appeared to suppress the virologic expression of cervical and urinary tract infection, whereas multiparity seemingly produced such an effect only in the cervix. Among both cervical and urinary excreters, a few shed virus thoughout pregnancy, and others shed virus intermittently; however, viral shedding most commonly began in late gestation and frequently continued into the postpartum period. Primary infection was not documented, and antibody status remained unchanged with the advent of viral excretion in most cases. Thus, reactivation of endogenous virus seems the most likely explanation for viral shedding in our population. Similar rates of isolation of Neisseria gonorrhoeae in excreters and nonexcreters further -argue against the other major possibility, venereal reinfection.
women remains incompletely defined. Increasing rates of active cervical infection with advancing gestation, reaching 11 %-28% in late pregnancy, have been demonstrated in prospective surveys of such dissimilar populations as Orientals and Navaho Indians as well as black American women of low socioeconomic status [1-3]. 'An adequate explanation of this phenomenon with regard to the types of infection involved has not appeared although reactivation and reinfection by the venereal route have been proposed [1-5]. The uniformly low rates of excretion (0%2%) observed during early gestation would seemingly predict similar rates among nonpregnant women. In fact, a recent study reported a 0% rate of active infection among females undergoing premarital or required gynecologic examinations [4]; however, in the same study,
Despite current interest in cervical cytomegalovirus (CMV) infection, the natural history of this infection in pregnant and nonpregnant Received for publication August 22, 1974, and in revised form December 13, 1974. This work was supported in part by research grant no. HD-01687 and training grant no. HD-143 from the National Institute of Child Health and Human Developm~nt, by research grant no. CA-13148 from the National Cancer Institute, by the National Foundation-March of Dimes, and by the Robert E. Meyer Foundation. We thank the nurses and other staff members of the Central Health Prenatal and Family Planning Clinics, Public Health Department, Birmingham, Alabama for their aid in this project; Mr. Rolfe Reynolds for technical assistance; and Ms. L. Gayle Arnold for assistance in the preparation of this manuscript. Please address requests for reprints to Dr. Sergio Stagno, UAB Medical Center, C. D.L. D. Building, Room 609, University Station, Birmingham, AJabama 35294. * Deceased.
522
Downloaded from http://jid.oxfordjournals.org/ at Thomas Jefferson University, Scott Library on March 11, 2015
From the Department of Pediatrics, School of Medicine. the University of Alabama in Birmingham; the Maternal and Child Health Division of the Jefferson County Health Department, Birmingham, Alabama; and the Section of Infectious Diseases, Perinatal Research Branch, National Institute of Neurological Diseases and Stroke, Bethesda, Maryland
Cervical Cytomegalovirus Excretion
Materials and Methods
From January 1971 through October 1972, 659 pregnant females of low socioeconomic status who attended one public health clinic in metropolitan Birmingham, Alabama, were studied prospectively and serially. At the time of the initial visit (mean, 17 weeks of .gestation) and again late in pregnancy (mean, 35 weeks), cervical swab and urine specimens as' well as blood samples were collected for ,:,iral isolation and CMV serologic testing. Women who shed virus at any time were recalled as soon as possible for an attempt at viral isolation from throat, white blood cells, and amniotic fluid when feasible. In this group, a mean of two additional cervical, urine, and serum samples were studied for evidence of viral infection during gesration.. Furthermore, throat swabs, colostrum', placenta, and newborns' urine specimens collected during the puerperium, as well as cervical swab specimens obtained four to six weeks postpartum, were examined for the presence of virus. From November 1972 through February 1973, single cervical swab and urine specimens were obtained for CMV isolation from 202 and 230 nonpregnant females, respectively. Both groups attended the same public health clinic as their pregnant cohorts, and those with previous conceptions were at least six months postpartum. Incidental to this study in both pregnant and
nonpregnant groups, the initial cervical swab specimens were studied for the presence of Neisseria gonorrhoeae by inoculation onto Thayer-Martin medium. Virology. Specimens were processed and examined as previously described; cell cultures containing monolayers of human embryonic skin fibroblasts were used [3]. CMV was identified by its characteristic CPE and by its failure to replicate in nonhuman and human epithelial cells. Serology. The antigen used for all serologic tests was the AD-169 strain of CMV. CF tests were performed on samples of serum obtained from all pregnant women at the Bureau of Laboratories of the Alabama State Health Department in Montgomery [6]. In addition, sera from selected women were analyzed for indirect HA [7] and indirect immunofluorescent (FA) antibodies. The latter assay is a modification of that described by Hanshaw [8]. Important differences between the two assays include the admixture of equal numbers of infected and uninfected cells in the antigen preparation, the use of highly purified antihuman IgG (Fc fragment) prepared from caprine sera in the laboratory of Dr. Alexander Lawton of this institution [9], an increase of incubation times from 30 to 60 min, and the application of Evans blue counterstain. The F A and indirect H A determinations were performed on sera from 41 and 26 randomly selected excreters and on sera from 141 and 56 subjects lacking CF antibodies, respectively. Titers of 1:8 were judged to be positive in the CF and indirect HA tests; a positive FA test at the 1: 16 dilution was taken as evidence of infection. Statistical analyses. Tests for statistical significance included X2 and Student's t-test. Results
Pregnant group. Among the 659 predominantly black, young pregnant females, cervical CMV excretion was demonstrated in 63 (9.5%). As measured by CF antibody, 52.7% of these women showed evidence of previous experience with CM V. Among those lacking C F antibodies, however, 58% and 76% were reactive
Downloaded from http://jid.oxfordjournals.org/ at Thomas Jefferson University, Scott Library on March 11, 2015
the rate of excretion reached 13% among nonpregnant women with suspected venereal disease. This finding suggested a possible venereal route of transmission, since viral excretion was related significantly only to a history of gonococcal infection. Some of the existing questions relevant to the nature of this infection can best be answered by a study of prevalence of cervical CMV infection in pregnant and nonpregnant women from the same population. We present here the results of such a study, which provide virologic and serologic data substantiating a phenomenon of suppression of viral activation early in pregnancy and further delineate the nature of this infection in the gravida.
523
524
Stagna et ai.
Table 1.
creased significantly from the first trimester (one of 194 or 0.5%) to the second and third trimesters of pregnancy (15 of 386 or 3.9% and 18 of 389 or 4.6%; P < 0.02). As with cervical shedding, the rate of viruria decreased progressively with advancing age. Multiparity, in contrast, had no influence on the shedding of virus in the urine (table 2). Active gonococcal infection was documented in 10.1% of the pregnant females. The rates were similar in the CMV excreters (eight of 63 or 12.7%) and nonexcreters (58 of 596 or 9.7%). Nonpregnant group. CMV was recovered from the cervix of 19 of 202 nonpregnant females (9.4%), a figure similar to that observed in late gestation among their pregnant cohorts. The age-specific rate of active cervical infection resembled closely that observed in the pregnant group; it remained fairly constant from 15 through 30 years of age. Virus was not isolated from the 18 women 30 years of age or older. Likewise, women who had had three or more pregnancies shed virus less frequently (one of 28) than did their less fecund cohorts (18 of 173), but the difference was not significant (P > 0.25). Active gonococcal infection was documented in 7.5% of the nonpregnant women. As in the pregnant group, the rate of N. 'gonorrhoeae isolation in CMV-excreting women was not different from that in nonexcretingwomen (two of 19 vs. 13 of 183). The use and type of contraception employed seemed to have. no influence on cervical CMV shedding, since among 103
Rates of isolation of cytomegalovirus from the cervix according to length of gestation. Trimester (%)
Source
First
Second
Third
Total*
Present studyt
1.6
6.1
11.3
63/659
Numazaki [I]
0
9.6
27.8
Montgomery [2}
2
7
12
23/153 (15) 14/176
(9.5)
(8)
Total no. with virus isolated/no. tested (%)*
4/256 (1.5)
34/504 (6.7)
65/481 (13.5)
100/988 (10. I)
* Total number with virus isolated/total number tested (percent). t Differences between first and second and those between second and third trimester rates were significant at P < 0.02. :j: Differences between first and second and those between second and third trimester rates were significant at P < 0.005 and P < 0.001, respectively.
Downloaded from http://jid.oxfordjournals.org/ at Thomas Jefferson University, Scott Library on March 11, 2015
in the indirect HA and FA tests, respectively. By extrapolation, then, 80.3% (indirect HA) and 88.8% (FA) of these young females had acquired CMV infection before they reached the childbearing years. Viral excretion (table 1) increased significantly with advancing gestation from 1.6% (three of 183) in the first trimester to 6.1% (22 of 359) and 11.3% (42 of 317) during the second and third trimesters, respectively. Two previous prospective surveys among Japanese and American women (the latter group composed of whites, blacks, and Navaho Indians) demonstrated a similar phenomenon [l, 2]. Summation of the data derived from these surveys and our own permitted the analysis of 988 pregnant females studied for evidence of viral infection at different gestational intervals (table 1). Active cervical CMV infection occurred in 10.1% of the populations studied. When isolation rates were related to length of gestation, a highly significant increase in virus shedding was found with advancing pregnancy (P < 0.005). In the present study, age-specific analysis demonstrated a relatively uniform rate of cervical excretion from 15 to 30 years . o f age (table 2). However, in the group less than 15 years of age, the rate of excretion was significantly higher than that in women over '30 (P < 0.05). A lower prevalence of genital infection was also noted in multiparous women (P < 0.05). CMV was recovered in the urine of 31 (4.7%) pregnant females, 14 of whom were found to be cervical excreters as well. Viruria also in-
525
Cervical Cytomegalovirus Excretion
Table 2. Recovery of cytomegalovirus from cervix and urine of pregnant women by age and parity. Group
Urine
5/30(17)* 41/372(11) 12/111(11) 4/41(10) 0/30(0)
3/38(8) 23/306(6) 5/123(4) 0/44(0) 0/29(0)
59/592(10) 1/67(2)
3/60(5) 28/540(5)
Age (years) '!":;14 15-20 21-25 26-30 ;;-:31
Parity 0-2 ;;-:3
* Number with virus isolated/total number tested (percent).
women taking oral contraceptives, 9.8% were actively infected, compared with 9.2% of 99 women using other methods or no contraception. Viruria was documented in six of 230 nonpregnant women (2.6%). Demographic data. Both pregnant and nonpregnant groups were drawn from 'a public health prenatal and family planning clinic caring mainly for indigent patients. Although minor dissimilarities existed among the three populations (table 3), intragroup, dem~graphic analysis revealed no significant differences between excreting and nonexcreting women with respect to race, age, marital status, and number of previous pregnancies. Pregnant cervical CMV excreter:s. Cervical Table 3. Demographic characteristics of pregnant and nonpregnant groups studied for cervical or urinary excretion of cytomegalovirus. Nonpregnant Characteristic
Pregnant
Cervical
Urinary
Age (years) Race Black White Marital status Married Single Previous pregnancies None
19.8 (5.1)
21.5 (5.8)
21.9 (5.5)
612 (93) 47 (7)
162 (81) 39 (19)
182 (79) 48 (21)
260 (39.5) 399 (60.5)
61 (30) 141 (70)
87 (38) 143 (62)
365 (55.7) 294 (44.3)
82 (40.5) 120 (59.5)
83 (36) 147 (64)
592 (89.9) 67 (10.1) 659
174 (86.6) 28 (13.4) 202
207 (90) 23 (10) 230
;;-:1 ,,;;;2 ;;-:3
Total
NOTE. Age is expressed as mean (±SD). All other values are expressed as number in given category (percent).
Downloaded from http://jid.oxfordjournals.org/ at Thomas Jefferson University, Scott Library on March 11, 2015
Cervix
CMV excretion was documented at least once during pregnancy in 63 women. Fifty-five were serially sampled; examples of their patterns of viral excretion are depicted in table 4. In 33, virus was recovered only in late pregnancy, and in 12 of these 33 excretion persisted into the postpartum period. Of the remainder, 13 shed virus only briefly during midgestation, while four and five excreted virus continuously and intermittently, respectively. The three patients who excreted CMV during the first trimester had the latter pattern; in two, shedding recurred only in the postpartum period, while in the other, CMV was reisolated in late gestation as well as after delivery. Simultaneous involvement of the urinary tract and mammary gland was common, as evidenced by the rates of virus isolation from urine (14 of 57 or 25%) and from colostrum (eight of 33 or 24%). In contrast, attempts to recover virus from 53, 27, and 40 samples of white blood cells, amniotic fluid, and placentas, respectively, were unsuccessful. One woman, excreting CMV in the third trimester, delivered a congenitally infected infant. Serologic evidence of primary infection was lacking among the excreters studied for at least four months after the recovery of virus (table 5). In only one patient was seroconversion documented by the CF test; however, this was not corroborated by either FA or indirect HA tests. A fourfold or greater rise in antibody levels occurred in six mothers, but only one had such a rise in all three serologic tests. Interestingly, four of the remainder had minimal antibody rises (fourfold) in only one of the tests. The vast majority of the infected women had stable antibody levels when first measured either before or at the time of initial CMV isolation. The low sensitivity of the CF tests, at least as employed in the present study, was further substantiated by the fact that 24% of the excreters lacked CF antibodies, whereas all subjects tested, including those without CF antibody, were reactive with the FA procedure, and all but one had measurable indirect HA antibodies. Among the women who had FA or indirect HA antibodies in their sera, cervical CMV excretion was documented in 14.9% and 15.3%, respectively.
Stagno et al.
526
Table 4.
Patterns of cervical excretion of cytomegalovirus during pregnancy and early postpartum period. Week of gestation Postpartum
Patterns *
15
20
25
30
35
I
+
2
+
3
+
4
+ +
5
2It
+
+ +
of virus; -
=
Total no. with pattern
12
13* 4
5
no isolation of virus; blank indicates that isolation was not attempted.
t Seven patients were not studied virologically during postpartum period.
* Five patients were not studied virologically during postpartum period. Discussion
Previous studies [1-3] have documented an impressively high rate of cervical and urinary tract infection with CMV during pregnancy. These data were generally interpreted to mean that pregnancy causes reactivation of latent virus in both sites. This interpretation was based on the belief that rates were much lower in normal adults. Furthermore, the populations of pregnant women studied were, in general, highly immune at the outset, and, in 'individual cases, antibody was most often present before viral shedding began. Primary involvement is, therefore, not a major contributor to this phenomenon. In prior studies unchanging levels of CF antibody before or after shedding from either the cervix or the urinary tract made reinfection by the respiratory route seem most unlikely. In addition to being an argument against venereal reinfection, the universal occurrence, as summarized here, of increasing incidence of viral excretion with advancing gestation led us to postulate that CMV was reactivated. Clearly, Table 5. Results of fluorescent antibody (F A), indi-rect HA (lHA), and CF tests of 59 pregnant women who excreted cytomegalovirus from the cervix. Result Seroconversion Rise in titer ;;:-::4-fold Stable titer before and after isolation of virus Stable titer at and after isolation of virus No antibody Total no. tested
FA
IHA
CF
0 2
0 3
1 3
23
15
20
16
7 1 26
21 14 59
0 41
however, the high rate of cervical excretion in late gestation in our population represents the base-line level when compared to the nonpregnant state. Rather than progressively activating latent cervical virus, pregnancy, in fact, appears to exert a suppressive effect on viral excretion in early gestation that becomes less pronounced with time, at least in women with prior immunity. The situation is less clear with excretion via the urinary tract. In this site, total excretion rates are lower than those for cervical excretion at any gestational interval, and the only significant increase is seen between very early pregnancy and the second trimester. Also, base-line values of urinary shedding in the nonpregnant population are intermediate (2.6%) and thus are not significantly different from the low value (0.5%) observed in the first trimester and the high value (4.6%) seen in late pregnancy. However, there is a trend toward suppression of urinary CMV excretion in the earliest stages of pregnancy that is seemingly active for shorter periods than that affecting the cervix. Proof of this suggestion awaits larger-scale studies. Examination of the patterns of viral shedding among cervical excreters indicates that suppression is not an all-or-none phenomenon, since first-trimester shedding was detectable in three of 10 excreters examined. In general, three patterns of cervical viral shedding were seen. Most commonly, it was detectable only in the third trimester and the postpartum period. A few women apparently excreted virus throughout pregnancy, while in a somewhat larger group, excretion was intermittent, occasionally including periods of shedding in early and late preg-
Downloaded from http://jid.oxfordjournals.org/ at Thomas Jefferson University, Scott Library on March 11, 2015
* Key: + = isolation
+
+
(1-2 months)
Cervical Cytomegalovirus Excretion
Irrespective of the type of recurrence, advancing age and multiparity appear to dampen significantly the virologic expression of genital infection, regardless of current pregnancy status. Similar results have been reported in studies of separate pregnant and nonpregnant populations [2, 4]. Viruria is also less common in older gravidas; however, fecundity appears to have little influence. These findings raise the possibility that advancing age may enhance latency of infection regardless of the site involved. Multiparity, on the other hand, would appear to exert its influence only in the cervix, an organ more directly influenced by the hormonal and mechanical changes attending the pregnant state. References 1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
Numazaki, Y., Yano, N., Morizuka, T., Takai, S., Ishida, N. Primary infection with human cytomegalovirus: virus isolation from healthy infants and pregnant women. Am. J. EpidemioI. 91:410-417, 1970. Montgomery, R., Youngblood, L., Medearis, D. N., Jr. Recovery of cytomegalovirus from the cervix in pregnancy. Pediatrics 49: 524-531, 1972. Reynolds, D. W., Stagno, S., Hosty, T. S., Tiller, M., Alford, C. A., Jr. Maternal cytomegalovirus excretion and perinatal infection. N. EngI. J. Med. 289:1-5, 1973. Jordan, M. C., Rousseau, W. E., Noble, G. R., Stewart, J. A., Chin, T. D. Y. Association of cervical cytomegaloviruses with venereal disease. N. Engl. J. Med. 288:932-934, 1973. Lang, D. J., Kummer, J. F. Demonstration of cytomegalovirus in semen. N. EngI. J. Med. 287:756-758, 1972. National Communicable Disease Center. Standard diagnostic complement fixation method and adaptation to micro test. Public Health Monograph no. 74, (USPHS publication no. 1228) U.S. Government Printing Office, Washington, D.C., 1965. Fuccillo, D. A., Moder, F. L., Traub, R. G., Hensen, S., Sever, J. L. Micro indirect hemagglutination test for cytomegalovirus. Appl. Microbiol. 21: 104-107, 1971. Hanshaw, J. B., Steinfield, H. J., White, C. J. FIourescent antibody test for cytomegalovirus macroglobulin. N. Engl. J. Med. 279:566-570, 1968. Cooper, M. D., Lawton, A. R., Bockman, D. E. Agammaglobulinaemia with b lymphocytes: specific defect of plasma-cell differentiation. Lancet. 2:791795, 1971. Solberg, D. A., Butler, J., Wagner, N. N. Sexual behavior in pregnancy. N. EngI. J. Med. 288: 10981103, 1973.
Downloaded from http://jid.oxfordjournals.org/ at Thomas Jefferson University, Scott Library on March 11, 2015
nancy. Serial testing for CF, indirect HA, and F A antibody levels failed to differentiate any of these groups. In fact, with all three procedures, serologic evidence of primary infection was lacking, and evidence for significant anamnestic response to reinfection was uncommonly encountered, and then only questionably. In the main, levels of all three antibodies remained stable with or without viral excretion. All of the cervical excreters had significant amounts of FA antibody when admitted to the study, and all but one possessed indirect HA antibody. Most of the excreters, then, had serologic evidence of prior exposure to CMV before the initiation of shedding; the same was true for the urinary excreters. However, a significant number lacked CF antibody on admission and failed to develop detectable amounts of this antibody despite active infection at these localized sites. Whether this finding represents insensitivity of the procedure as employed here or a reduced antigenic load in these patients remains to be determined. Thus this extraordinary prevalence of cervical CMV excretion in both pregnant· and nonpregnant women must, at least in this population, represent recurrent infection. Whether these recurrences are due to reactivation of endogenous virus, venereal reinfection, or both cannot be conclusively determined at presenL However, our findings of stable antibody levels, similar rates of gonorrhea in excreters and nonexcreters, and the increasing rate of excretion during pregnancy in the face of a documented tendency toward decreasing coital activity during this period [10] suggest that the former type of infection is the major contributor, at least in highly immune populations. Similarly, productive infection of the urinary tract and mammary gland probably represents predominantly local reactivation, based on unchanging humoral immunity and the failure to isolate virus from sites other than the cervix. The possibility of reinfection with a heterologous strain of virus of such dissimilar antigenic composition as to preclude a measurable antibody response to the prototype strain obviously cannot be excluded. For this reason, definition of the true incidence or, in fact, the very existence of reinfection will be extraordinarily difficult to establish.
527
MAY 1975
Cervical Cytomegalovirus Excretion in Pregnant and Nonpregnant Women: Suppression in Early Gestation Sergio Stagno, David Reynolds, Alex Tsiantos, * David A. Fuccillo, Richard Smith, Mary Tiller, and Charles A. Alford, Jr.
Comparison of 659 pregnant and 202 nonpregnant women with similar demographic characteristics showed overall rates of cervical cytomegalovirus excretion that were identical (9.5% vs. 9.4%) and were surprisingly high, especially since 89% of the pregnant group possessed antibody to cytomegalovirus when admitted to the study. Prevalence of cytomegalovirus among gravidas was significantly lower during the first (1.6%) than during the third (11.3%) trimester. Thus, early pregnancy appeared to exert a suppressive effect on viral excretion that waned with advancing gestation. A similar but less significant occurrence was observed in the two groups with respect to viruria. Increasing age also appeared to suppress the virologic expression of cervical and urinary tract infection, whereas multiparity seemingly produced such an effect only in the cervix. Among both cervical and urinary excreters, a few shed virus thoughout pregnancy, and others shed virus intermittently; however, viral shedding most commonly began in late gestation and frequently continued into the postpartum period. Primary infection was not documented, and antibody status remained unchanged with the advent of viral excretion in most cases. Thus, reactivation of endogenous virus seems the most likely explanation for viral shedding in our population. Similar rates of isolation of Neisseria gonorrhoeae in excreters and nonexcreters further -argue against the other major possibility, venereal reinfection.
women remains incompletely defined. Increasing rates of active cervical infection with advancing gestation, reaching 11 %-28% in late pregnancy, have been demonstrated in prospective surveys of such dissimilar populations as Orientals and Navaho Indians as well as black American women of low socioeconomic status [1-3]. 'An adequate explanation of this phenomenon with regard to the types of infection involved has not appeared although reactivation and reinfection by the venereal route have been proposed [1-5]. The uniformly low rates of excretion (0%2%) observed during early gestation would seemingly predict similar rates among nonpregnant women. In fact, a recent study reported a 0% rate of active infection among females undergoing premarital or required gynecologic examinations [4]; however, in the same study,
Despite current interest in cervical cytomegalovirus (CMV) infection, the natural history of this infection in pregnant and nonpregnant Received for publication August 22, 1974, and in revised form December 13, 1974. This work was supported in part by research grant no. HD-01687 and training grant no. HD-143 from the National Institute of Child Health and Human Developm~nt, by research grant no. CA-13148 from the National Cancer Institute, by the National Foundation-March of Dimes, and by the Robert E. Meyer Foundation. We thank the nurses and other staff members of the Central Health Prenatal and Family Planning Clinics, Public Health Department, Birmingham, Alabama for their aid in this project; Mr. Rolfe Reynolds for technical assistance; and Ms. L. Gayle Arnold for assistance in the preparation of this manuscript. Please address requests for reprints to Dr. Sergio Stagno, UAB Medical Center, C. D.L. D. Building, Room 609, University Station, Birmingham, AJabama 35294. * Deceased.
522
Downloaded from http://jid.oxfordjournals.org/ at Thomas Jefferson University, Scott Library on March 11, 2015
From the Department of Pediatrics, School of Medicine. the University of Alabama in Birmingham; the Maternal and Child Health Division of the Jefferson County Health Department, Birmingham, Alabama; and the Section of Infectious Diseases, Perinatal Research Branch, National Institute of Neurological Diseases and Stroke, Bethesda, Maryland
Cervical Cytomegalovirus Excretion
Materials and Methods
From January 1971 through October 1972, 659 pregnant females of low socioeconomic status who attended one public health clinic in metropolitan Birmingham, Alabama, were studied prospectively and serially. At the time of the initial visit (mean, 17 weeks of .gestation) and again late in pregnancy (mean, 35 weeks), cervical swab and urine specimens as' well as blood samples were collected for ,:,iral isolation and CMV serologic testing. Women who shed virus at any time were recalled as soon as possible for an attempt at viral isolation from throat, white blood cells, and amniotic fluid when feasible. In this group, a mean of two additional cervical, urine, and serum samples were studied for evidence of viral infection during gesration.. Furthermore, throat swabs, colostrum', placenta, and newborns' urine specimens collected during the puerperium, as well as cervical swab specimens obtained four to six weeks postpartum, were examined for the presence of virus. From November 1972 through February 1973, single cervical swab and urine specimens were obtained for CMV isolation from 202 and 230 nonpregnant females, respectively. Both groups attended the same public health clinic as their pregnant cohorts, and those with previous conceptions were at least six months postpartum. Incidental to this study in both pregnant and
nonpregnant groups, the initial cervical swab specimens were studied for the presence of Neisseria gonorrhoeae by inoculation onto Thayer-Martin medium. Virology. Specimens were processed and examined as previously described; cell cultures containing monolayers of human embryonic skin fibroblasts were used [3]. CMV was identified by its characteristic CPE and by its failure to replicate in nonhuman and human epithelial cells. Serology. The antigen used for all serologic tests was the AD-169 strain of CMV. CF tests were performed on samples of serum obtained from all pregnant women at the Bureau of Laboratories of the Alabama State Health Department in Montgomery [6]. In addition, sera from selected women were analyzed for indirect HA [7] and indirect immunofluorescent (FA) antibodies. The latter assay is a modification of that described by Hanshaw [8]. Important differences between the two assays include the admixture of equal numbers of infected and uninfected cells in the antigen preparation, the use of highly purified antihuman IgG (Fc fragment) prepared from caprine sera in the laboratory of Dr. Alexander Lawton of this institution [9], an increase of incubation times from 30 to 60 min, and the application of Evans blue counterstain. The F A and indirect H A determinations were performed on sera from 41 and 26 randomly selected excreters and on sera from 141 and 56 subjects lacking CF antibodies, respectively. Titers of 1:8 were judged to be positive in the CF and indirect HA tests; a positive FA test at the 1: 16 dilution was taken as evidence of infection. Statistical analyses. Tests for statistical significance included X2 and Student's t-test. Results
Pregnant group. Among the 659 predominantly black, young pregnant females, cervical CMV excretion was demonstrated in 63 (9.5%). As measured by CF antibody, 52.7% of these women showed evidence of previous experience with CM V. Among those lacking C F antibodies, however, 58% and 76% were reactive
Downloaded from http://jid.oxfordjournals.org/ at Thomas Jefferson University, Scott Library on March 11, 2015
the rate of excretion reached 13% among nonpregnant women with suspected venereal disease. This finding suggested a possible venereal route of transmission, since viral excretion was related significantly only to a history of gonococcal infection. Some of the existing questions relevant to the nature of this infection can best be answered by a study of prevalence of cervical CMV infection in pregnant and nonpregnant women from the same population. We present here the results of such a study, which provide virologic and serologic data substantiating a phenomenon of suppression of viral activation early in pregnancy and further delineate the nature of this infection in the gravida.
523
524
Stagna et ai.
Table 1.
creased significantly from the first trimester (one of 194 or 0.5%) to the second and third trimesters of pregnancy (15 of 386 or 3.9% and 18 of 389 or 4.6%; P < 0.02). As with cervical shedding, the rate of viruria decreased progressively with advancing age. Multiparity, in contrast, had no influence on the shedding of virus in the urine (table 2). Active gonococcal infection was documented in 10.1% of the pregnant females. The rates were similar in the CMV excreters (eight of 63 or 12.7%) and nonexcreters (58 of 596 or 9.7%). Nonpregnant group. CMV was recovered from the cervix of 19 of 202 nonpregnant females (9.4%), a figure similar to that observed in late gestation among their pregnant cohorts. The age-specific rate of active cervical infection resembled closely that observed in the pregnant group; it remained fairly constant from 15 through 30 years of age. Virus was not isolated from the 18 women 30 years of age or older. Likewise, women who had had three or more pregnancies shed virus less frequently (one of 28) than did their less fecund cohorts (18 of 173), but the difference was not significant (P > 0.25). Active gonococcal infection was documented in 7.5% of the nonpregnant women. As in the pregnant group, the rate of N. 'gonorrhoeae isolation in CMV-excreting women was not different from that in nonexcretingwomen (two of 19 vs. 13 of 183). The use and type of contraception employed seemed to have. no influence on cervical CMV shedding, since among 103
Rates of isolation of cytomegalovirus from the cervix according to length of gestation. Trimester (%)
Source
First
Second
Third
Total*
Present studyt
1.6
6.1
11.3
63/659
Numazaki [I]
0
9.6
27.8
Montgomery [2}
2
7
12
23/153 (15) 14/176
(9.5)
(8)
Total no. with virus isolated/no. tested (%)*
4/256 (1.5)
34/504 (6.7)
65/481 (13.5)
100/988 (10. I)
* Total number with virus isolated/total number tested (percent). t Differences between first and second and those between second and third trimester rates were significant at P < 0.02. :j: Differences between first and second and those between second and third trimester rates were significant at P < 0.005 and P < 0.001, respectively.
Downloaded from http://jid.oxfordjournals.org/ at Thomas Jefferson University, Scott Library on March 11, 2015
in the indirect HA and FA tests, respectively. By extrapolation, then, 80.3% (indirect HA) and 88.8% (FA) of these young females had acquired CMV infection before they reached the childbearing years. Viral excretion (table 1) increased significantly with advancing gestation from 1.6% (three of 183) in the first trimester to 6.1% (22 of 359) and 11.3% (42 of 317) during the second and third trimesters, respectively. Two previous prospective surveys among Japanese and American women (the latter group composed of whites, blacks, and Navaho Indians) demonstrated a similar phenomenon [l, 2]. Summation of the data derived from these surveys and our own permitted the analysis of 988 pregnant females studied for evidence of viral infection at different gestational intervals (table 1). Active cervical CMV infection occurred in 10.1% of the populations studied. When isolation rates were related to length of gestation, a highly significant increase in virus shedding was found with advancing pregnancy (P < 0.005). In the present study, age-specific analysis demonstrated a relatively uniform rate of cervical excretion from 15 to 30 years . o f age (table 2). However, in the group less than 15 years of age, the rate of excretion was significantly higher than that in women over '30 (P < 0.05). A lower prevalence of genital infection was also noted in multiparous women (P < 0.05). CMV was recovered in the urine of 31 (4.7%) pregnant females, 14 of whom were found to be cervical excreters as well. Viruria also in-
525
Cervical Cytomegalovirus Excretion
Table 2. Recovery of cytomegalovirus from cervix and urine of pregnant women by age and parity. Group
Urine
5/30(17)* 41/372(11) 12/111(11) 4/41(10) 0/30(0)
3/38(8) 23/306(6) 5/123(4) 0/44(0) 0/29(0)
59/592(10) 1/67(2)
3/60(5) 28/540(5)
Age (years) '!":;14 15-20 21-25 26-30 ;;-:31
Parity 0-2 ;;-:3
* Number with virus isolated/total number tested (percent).
women taking oral contraceptives, 9.8% were actively infected, compared with 9.2% of 99 women using other methods or no contraception. Viruria was documented in six of 230 nonpregnant women (2.6%). Demographic data. Both pregnant and nonpregnant groups were drawn from 'a public health prenatal and family planning clinic caring mainly for indigent patients. Although minor dissimilarities existed among the three populations (table 3), intragroup, dem~graphic analysis revealed no significant differences between excreting and nonexcreting women with respect to race, age, marital status, and number of previous pregnancies. Pregnant cervical CMV excreter:s. Cervical Table 3. Demographic characteristics of pregnant and nonpregnant groups studied for cervical or urinary excretion of cytomegalovirus. Nonpregnant Characteristic
Pregnant
Cervical
Urinary
Age (years) Race Black White Marital status Married Single Previous pregnancies None
19.8 (5.1)
21.5 (5.8)
21.9 (5.5)
612 (93) 47 (7)
162 (81) 39 (19)
182 (79) 48 (21)
260 (39.5) 399 (60.5)
61 (30) 141 (70)
87 (38) 143 (62)
365 (55.7) 294 (44.3)
82 (40.5) 120 (59.5)
83 (36) 147 (64)
592 (89.9) 67 (10.1) 659
174 (86.6) 28 (13.4) 202
207 (90) 23 (10) 230
;;-:1 ,,;;;2 ;;-:3
Total
NOTE. Age is expressed as mean (±SD). All other values are expressed as number in given category (percent).
Downloaded from http://jid.oxfordjournals.org/ at Thomas Jefferson University, Scott Library on March 11, 2015
Cervix
CMV excretion was documented at least once during pregnancy in 63 women. Fifty-five were serially sampled; examples of their patterns of viral excretion are depicted in table 4. In 33, virus was recovered only in late pregnancy, and in 12 of these 33 excretion persisted into the postpartum period. Of the remainder, 13 shed virus only briefly during midgestation, while four and five excreted virus continuously and intermittently, respectively. The three patients who excreted CMV during the first trimester had the latter pattern; in two, shedding recurred only in the postpartum period, while in the other, CMV was reisolated in late gestation as well as after delivery. Simultaneous involvement of the urinary tract and mammary gland was common, as evidenced by the rates of virus isolation from urine (14 of 57 or 25%) and from colostrum (eight of 33 or 24%). In contrast, attempts to recover virus from 53, 27, and 40 samples of white blood cells, amniotic fluid, and placentas, respectively, were unsuccessful. One woman, excreting CMV in the third trimester, delivered a congenitally infected infant. Serologic evidence of primary infection was lacking among the excreters studied for at least four months after the recovery of virus (table 5). In only one patient was seroconversion documented by the CF test; however, this was not corroborated by either FA or indirect HA tests. A fourfold or greater rise in antibody levels occurred in six mothers, but only one had such a rise in all three serologic tests. Interestingly, four of the remainder had minimal antibody rises (fourfold) in only one of the tests. The vast majority of the infected women had stable antibody levels when first measured either before or at the time of initial CMV isolation. The low sensitivity of the CF tests, at least as employed in the present study, was further substantiated by the fact that 24% of the excreters lacked CF antibodies, whereas all subjects tested, including those without CF antibody, were reactive with the FA procedure, and all but one had measurable indirect HA antibodies. Among the women who had FA or indirect HA antibodies in their sera, cervical CMV excretion was documented in 14.9% and 15.3%, respectively.
Stagno et al.
526
Table 4.
Patterns of cervical excretion of cytomegalovirus during pregnancy and early postpartum period. Week of gestation Postpartum
Patterns *
15
20
25
30
35
I
+
2
+
3
+
4
+ +
5
2It
+
+ +
of virus; -
=
Total no. with pattern
12
13* 4
5
no isolation of virus; blank indicates that isolation was not attempted.
t Seven patients were not studied virologically during postpartum period.
* Five patients were not studied virologically during postpartum period. Discussion
Previous studies [1-3] have documented an impressively high rate of cervical and urinary tract infection with CMV during pregnancy. These data were generally interpreted to mean that pregnancy causes reactivation of latent virus in both sites. This interpretation was based on the belief that rates were much lower in normal adults. Furthermore, the populations of pregnant women studied were, in general, highly immune at the outset, and, in 'individual cases, antibody was most often present before viral shedding began. Primary involvement is, therefore, not a major contributor to this phenomenon. In prior studies unchanging levels of CF antibody before or after shedding from either the cervix or the urinary tract made reinfection by the respiratory route seem most unlikely. In addition to being an argument against venereal reinfection, the universal occurrence, as summarized here, of increasing incidence of viral excretion with advancing gestation led us to postulate that CMV was reactivated. Clearly, Table 5. Results of fluorescent antibody (F A), indi-rect HA (lHA), and CF tests of 59 pregnant women who excreted cytomegalovirus from the cervix. Result Seroconversion Rise in titer ;;:-::4-fold Stable titer before and after isolation of virus Stable titer at and after isolation of virus No antibody Total no. tested
FA
IHA
CF
0 2
0 3
1 3
23
15
20
16
7 1 26
21 14 59
0 41
however, the high rate of cervical excretion in late gestation in our population represents the base-line level when compared to the nonpregnant state. Rather than progressively activating latent cervical virus, pregnancy, in fact, appears to exert a suppressive effect on viral excretion in early gestation that becomes less pronounced with time, at least in women with prior immunity. The situation is less clear with excretion via the urinary tract. In this site, total excretion rates are lower than those for cervical excretion at any gestational interval, and the only significant increase is seen between very early pregnancy and the second trimester. Also, base-line values of urinary shedding in the nonpregnant population are intermediate (2.6%) and thus are not significantly different from the low value (0.5%) observed in the first trimester and the high value (4.6%) seen in late pregnancy. However, there is a trend toward suppression of urinary CMV excretion in the earliest stages of pregnancy that is seemingly active for shorter periods than that affecting the cervix. Proof of this suggestion awaits larger-scale studies. Examination of the patterns of viral shedding among cervical excreters indicates that suppression is not an all-or-none phenomenon, since first-trimester shedding was detectable in three of 10 excreters examined. In general, three patterns of cervical viral shedding were seen. Most commonly, it was detectable only in the third trimester and the postpartum period. A few women apparently excreted virus throughout pregnancy, while in a somewhat larger group, excretion was intermittent, occasionally including periods of shedding in early and late preg-
Downloaded from http://jid.oxfordjournals.org/ at Thomas Jefferson University, Scott Library on March 11, 2015
* Key: + = isolation
+
+
(1-2 months)
Cervical Cytomegalovirus Excretion
Irrespective of the type of recurrence, advancing age and multiparity appear to dampen significantly the virologic expression of genital infection, regardless of current pregnancy status. Similar results have been reported in studies of separate pregnant and nonpregnant populations [2, 4]. Viruria is also less common in older gravidas; however, fecundity appears to have little influence. These findings raise the possibility that advancing age may enhance latency of infection regardless of the site involved. Multiparity, on the other hand, would appear to exert its influence only in the cervix, an organ more directly influenced by the hormonal and mechanical changes attending the pregnant state. References 1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
Numazaki, Y., Yano, N., Morizuka, T., Takai, S., Ishida, N. Primary infection with human cytomegalovirus: virus isolation from healthy infants and pregnant women. Am. J. EpidemioI. 91:410-417, 1970. Montgomery, R., Youngblood, L., Medearis, D. N., Jr. Recovery of cytomegalovirus from the cervix in pregnancy. Pediatrics 49: 524-531, 1972. Reynolds, D. W., Stagno, S., Hosty, T. S., Tiller, M., Alford, C. A., Jr. Maternal cytomegalovirus excretion and perinatal infection. N. EngI. J. Med. 289:1-5, 1973. Jordan, M. C., Rousseau, W. E., Noble, G. R., Stewart, J. A., Chin, T. D. Y. Association of cervical cytomegaloviruses with venereal disease. N. Engl. J. Med. 288:932-934, 1973. Lang, D. J., Kummer, J. F. Demonstration of cytomegalovirus in semen. N. EngI. J. Med. 287:756-758, 1972. National Communicable Disease Center. Standard diagnostic complement fixation method and adaptation to micro test. Public Health Monograph no. 74, (USPHS publication no. 1228) U.S. Government Printing Office, Washington, D.C., 1965. Fuccillo, D. A., Moder, F. L., Traub, R. G., Hensen, S., Sever, J. L. Micro indirect hemagglutination test for cytomegalovirus. Appl. Microbiol. 21: 104-107, 1971. Hanshaw, J. B., Steinfield, H. J., White, C. J. FIourescent antibody test for cytomegalovirus macroglobulin. N. Engl. J. Med. 279:566-570, 1968. Cooper, M. D., Lawton, A. R., Bockman, D. E. Agammaglobulinaemia with b lymphocytes: specific defect of plasma-cell differentiation. Lancet. 2:791795, 1971. Solberg, D. A., Butler, J., Wagner, N. N. Sexual behavior in pregnancy. N. EngI. J. Med. 288: 10981103, 1973.
Downloaded from http://jid.oxfordjournals.org/ at Thomas Jefferson University, Scott Library on March 11, 2015
nancy. Serial testing for CF, indirect HA, and F A antibody levels failed to differentiate any of these groups. In fact, with all three procedures, serologic evidence of primary infection was lacking, and evidence for significant anamnestic response to reinfection was uncommonly encountered, and then only questionably. In the main, levels of all three antibodies remained stable with or without viral excretion. All of the cervical excreters had significant amounts of FA antibody when admitted to the study, and all but one possessed indirect HA antibody. Most of the excreters, then, had serologic evidence of prior exposure to CMV before the initiation of shedding; the same was true for the urinary excreters. However, a significant number lacked CF antibody on admission and failed to develop detectable amounts of this antibody despite active infection at these localized sites. Whether this finding represents insensitivity of the procedure as employed here or a reduced antigenic load in these patients remains to be determined. Thus this extraordinary prevalence of cervical CMV excretion in both pregnant· and nonpregnant women must, at least in this population, represent recurrent infection. Whether these recurrences are due to reactivation of endogenous virus, venereal reinfection, or both cannot be conclusively determined at presenL However, our findings of stable antibody levels, similar rates of gonorrhea in excreters and nonexcreters, and the increasing rate of excretion during pregnancy in the face of a documented tendency toward decreasing coital activity during this period [10] suggest that the former type of infection is the major contributor, at least in highly immune populations. Similarly, productive infection of the urinary tract and mammary gland probably represents predominantly local reactivation, based on unchanging humoral immunity and the failure to isolate virus from sites other than the cervix. The possibility of reinfection with a heterologous strain of virus of such dissimilar antigenic composition as to preclude a measurable antibody response to the prototype strain obviously cannot be excluded. For this reason, definition of the true incidence or, in fact, the very existence of reinfection will be extraordinarily difficult to establish.
527
