key: cord-008499-tl3i7uzd
authors: Robb, James A.; Benirschke, Kurt; Barmeyer, Robert
title: Intrauterine latent herpes simplex virus infection(): I. Spontaneous abortion
date: 2007-11-06
journal: Hum Pathol
DOI: 10.1016/s0046-8177(86)80561-5
sha: 
doc_id: 8499
cord_uid: tl3i7uzd

Herpes simplex virus (HSV, probably type 2) antigen was detected in nonpregnant and pregnant endometria, placentae, umbilical cords, and neonatal tissues (companion paper) by avidinbiotin complex immunohistochemical studies. HSV cytologic abnormalities were not detected in any of the 380 cases examined: included were specimens from therapeutic and spontaneous abortions (200 cases) and endometrial curettage (180 cases). The presence of inflammation was not correlated with HSV positivity. Endometrial HSV positivity was significantly correlated with normal late secretory phase (40 per cent of specimens positive), abnormal secretory phase (67 per cent positive), and therapeutic (33 per cent positive) versus spontaneous (26 per cent positive) abortions. Placental HSV positivity was significantly correlated with spontaneous (39 per cent positive) versus therapeutic (14 per cent positive) abortions and with blighted ova (67 per cent positive). No significant correlation was found between HSV positivity and a clinical history of oral or genital HSV infection in either the patient or the male partner. The data support the concept of a subclinical latent intrauterine endometrial HSV infection that is hormonally regulated and can produce transplacental infection of the embryo or fetus, with variable consequences.

Acute intrauterine herpes simplex virus (HSV) infection, although rare, t has been increasing in frequency. 2 Our stndy was prompted by the following case of acute intrauterine HSV infection. A full-term infant was delivered alive with acute HSV infection of the skin and died a few days later of systemic infection. The presence of acute HSV infection in the skin and brain was confirmed by morphologic HSVspecific cytologic studies, culture, and transmission electron microscopy. No evidence of HSV infection was found in routine studies of the placenta and umbilical cord by the same techniques. Transplacental infection, however, must have occurred, because the infection was not acquired from the birth canal during delivery. A newly developed and highly sensitive glucose oxidase-avidin-biotin immunohistochemical technique for formalin-fixed tissues 3 demonstrated HSV-specific antigen ill phagocytic cells of the placental subamnionic chorion, in Hofbauer cells in chorionic villi, and in the subamnionic mesenchyme of the umbilical cord. No inflammation or HSV cytologic abnormalities were present in these areas.

Spontaneous and therapeutic abortion material, endometrial curettage tissue, full-term placentae and umbilical cords, and stillborn and liveborn neonatal tissues were studied by HSV-specific immunohistochemical methods to evaluate the prevalence and distribution of these HSV antigens in cases in which HSV infection was not suspected or detectable as HSV cytologic abnormalities by light microscopy, virus particles by transmission electron microscopy, or infectious HSV by direct cultnre. The presence and tissne distribution of the HSV antigens were then correlated with the finding for normal and pathologic conditions. The data concerning the fullterm neonates are presented in the companion paper. 4 The data concerning the detection of persistent intranterine HSV infection and its correlation with pathologic states in abortion and curettage material are presented below.

All tissues were routine stn'gical (products of conception, endometrial curettage, placentae, and umbilical cords) or autopsy (neonates and infants) specimens fixed in 10 per cent buffered formalin or Bouin's solution (placentae and cords only) and embedded in paraffin. The products of conception and endometrial tissues had been collected from 1981 to 1984 at the Green Hospital of Scripps Clinic. The placentae, umbilical cords, and neonatal or infant tissues had been collected from 1981 to 1984 at the University of California San Diego Medical Center. Some neonatal cases had been submitted from other hospitals for consultation. Serial 4-I.tm sections were attached to glass slides with either 1 per cent Elmer's glue in deionized water or chromic acidfpoly-D-Iysine (50 per cent chromic acid in deionized water for 30 minntes at room temperature, followed by rinsing and 0.01 per cent poly-D-lysine [Sigma, St. Louis, Missouri] in deionized water for 30 minutes at room temperature, followed by rinsing). The paraffin sections were not dried in an oven, but were left at room temperature for at least two hours before routine rentoval of paraffin and immunohistochemical staining.

All of the following steps were done at room temperature in a humidified box. The tissue sections were treated with 0.25 per cent crude porcine trypsin (Sigma, catalog no. T-8128) in phosphate-buffered saline (PBS, pH 7.4) for 60 minutes and rinsed with PBS by a jet-dribble technique. Primary polyclonal rabbit anti-HSV type 1 and type 2 IgG (Dako, Santa Barbara, California; lots 059A and 1188, respectively) and nonimmune rabbit IgG (Cappell Labs, West Chester, Pennsylvania) were incubated for 22 to 24 hours at 1:2,000 (HSV types 1 and 2) and at 10 txg/ml (nonimmune rabbit IgG) in PBS+ (PBS with 0.2 per cent sodium azide [Sigma] and 0.1 per cent bovine serum albumin [Calbiochem, La Jolla, California; type V]). The tissues were again rinsed with PBS, and biotinylated goat antirabbit IgG (3.25 lag/ntl PBS + ; Vector Labs, Burlingame, California), biotinylated horse antimouse IgG (30 Ixg/ml, Vector), or biotinylated goat anti-guinea pig IgG (7.5 Ixg/ml, Vector), depending on the species of primary antibody (see table 2), was added for 30 minutes. After PBS rinsing, a glucose oxidase-avidin-biotin complex (ABC-GO, Vector) was added for 30 minutes at room temperature. The staining reaction was identical to the previously described glucose oxidaseavidin-biotin (GAB) technique originally used in this study. 3 Tetranitroblue tetrazolium (TNBT) formazan formation (Vector) was used to detect sites of antibody binding. This "colorization reaction" was identical to that originally used in this study. 3 After a deionized water rinse, aqueous counterstaining was accomplished with 0.25 per cent metanil yellow in deionized water for 1 minute, deionized water rinse, and 0.1 per cent nuclear fast red for 5 minutes; a 0.45-~m filter (Gelman Acrodisc, Gelman Sciences, Ann Arbor, Michigan) was used at the time of addition to remove microcrystals. Aquamount (Lerner Labs, New Haven, Connecticut) was used for coverslipping. No loss of stain occurred during a four-year period.

For statistical analyses a Clinfo software program was used in the Scripps Clinic General Clinical Research Center. All data were nonparametric, except for the patient age data, which were normally distributed and analyzed by the Anderson-Darling t-test. The nonparametric data were analyzed with the Wilcoxon nonpaired rank sum test.

The rabbit anti-HSV 2 antibody staining in placental tissues was cytoplasmic and occurred in individual cells in the subamnionic chorion ( fig. 1 ) and in single Hofbauer cells in chorionic villi ( fig. 2 ). Umbil-ical cord staining occurred in single cells in tile subamnionic mesenchyme ( fig. 3 ), perivascular mesenchyme ( fig. 4 ) and, rarely, vascular smooth muscle ( fig. 4 ). No nuclear, amnionic, trophoblastic, or endothelial staining was detected in placental tissues. Apical and/or diffuse cytoplasmic staining occurred focally in the endometrial epithelium in both products of conception and nonpregnant endometrium, without specific nuclear staining ( fig. 5 ). Single choriodecidual cells were positive, as in the placentae ( fig. 6 ). Endothelial staining was rare (three of 380 cases) and was fotmd only in choriodecidual tissue, where it was both cytoplasmic and nuclear ( fig. 6 ).

Acute HSV endometrial epithelial infections, although rare, have been documented. 5,6 Altered endometrial epithelial nuclei with features suggestive of HSV infection, but without viral particles by transmission electron microscopy, were found in 7 per cent of a group of 200 cases in which endometrium was associated with trophoblastic tissue. 7 Endothelial cells can be infected by HSV, with subsequent altered function and death. 8,9 Histologically occult cytomegalovirus (CMV) infection has been detected in human tissue by immunohistochemical methods I° and by in situ DNA hybridization. II Either the detected antigens or their detecting antibodies were at a low concentration, because the screening anti-HSV 2 antibody titer difference between acute and chronic or latent infection was about 100-fold (acute--1:20,000 for two hours; chronic or latent--1:2,000 for 20 hours). The pattern of embryonic and neonatal staining is described in the companion paper. 4

Other virus-infected tissues. Table 1 demonstrates the specificity of the anti-HSV antibodies when tested against culture-proven acute HSV 1 infection in the skin ( fig. 7 ), brain, and esophagus, and against HSV 2 infection in the liver (acute) and adrenal gland (subacute, i.e., resolving intrauterine infection, several weeks after acute infection; fig. 7 ). Anti-HSV 1' and HSV 2 antibodies strongly cross-react in tissues acutely infected with HSV 1 and HSV 2 (see table 1 for end-point titers). The antibodies were absolutely negative in other tissues with culture-proven and/or immunohistochemically proven infections with the other three herpetoviruses (herpes zoster virus, CMV, and Epstein-Barr virus) and other miscellaneous viruses (hepatitis A and B, papillomavirus, and echovirus). A placenta with culture-proven listeria infection was also negative. The virus used as antigen for antibody production was purified from culture medium containing fetal bovine serum. For that reason, the antibodies were tested against tissues containing both human and bovine alpha-fetoprotein and were found to be negative (table 1) . Furthermore, adsorption of the anti-HSV 2 IgG with fresh fetal bovine liver powder (acetone-extracted) did not decrease the specific staining in target tissues. Cnlt n re/cytopat hology Cytopathology Cnltn re/c)'topathology Cnhure/cytopathology * HSV 1 and ttSV 2 antibody tilers against tile tlSV 1 acutely infected skin were I:16,000 and 1:8,000, respectively; tile liters against the HSV 2 subacutely infected adrenal gland were 1:4,000 and 1:!6,000, respectively. Other antiviral antibodies. Table 2 lists the staining results for various HSV-specific and nonspecific antibodies with the various target tissues. All of the UCSD/DR rabbit antisera and the Yale/GDH guinea pig antisera were coded samples at the time of tissue staining and evaluation. T h e following findings strongly support our hypothesis that the staining detected in the target and patient tissues with the Dako rabbit anti-HSV 2 IgG (lot 1188, our screening antibody) is specific for HSV and, probably, for type 2 infection in t h e products of conception, placentae, umbilical cords, and endometrial curettage specimens:

1. All preinfection, noninfected, and non-HSV sera were negative with all target tissues, with one exception. One of 11 UCSD/DR HSV 1 antisera produced 4 + staining of the vascular smooth muscle of the umbilical cord target ( fig. 4 ). The meaning of this finding is not known, although histologically occult CMV infection has been detected in human smooth muscle by immunohistochemical methods, l° 2. All HSV-infected animal sera were positive with at least one target tissue. There was wide individual variation, however, when the same target was used with different sera and different targets were used with the same sera. This phenomenon of variation in individual animal antibody expression to individual virus antigens is to be expected, both in humans 12 and in syngeneic mouse littermates, r3 as observed when cloned virus was used for "identical infection." One example o f the specificity o f the rabbit anti-HSV 2 antibody is shown in figure 8 . This gastric biopsy specimen from a patient with the acquired immunodeficiency syndrome shows CMVpositive cells (goat anti-CMV, Polysciences, 1:200 with trypsin pretreatment), which are negative for HSV 2 antigen, while adjacent gastrothelial cells show latent HSV 2 infection.

3. All of the NIH/MZ mouse anti-HSV monoclonal antibodies were positive in at least three targets. These antisera stain cells infected acutely with both HSV 1 and HSV 2 in cell culture.

4. The NIH/MZ mouse monoclonal antibody ICP4 recognizes an immediate/early (alpha) antigen in a nuclear location in both HSV 1 and 2 acute infections in cell culture. Scattered anmionic nuclei in the placental target were unequivocally positive with this antibody ( fig. 9 ), an indication of latent infection in the amnion. Intranuclear staining by ICP4 was also detected in chorionic nuclei in one case ( fig. 9 ). Negative control antibodies (negative in all above target tissues): Rabbit--normal nonimmune IgG, anti-fetuin, anti-fetal bovine serum, anti-callous keratin, anti-carcinoembryonic antigen, and auti-EBV. Antibovine alpha-fetoprotein was positive only in phagocytic cells of SAC, a subset of which were HSV-2-positive by dual labeling experiments. Guinea pig--normal nonimmune IgG, anti-insulin, and anti-varicella/zoster. Mouse--MOPC-21 and anticytokeratius 35BHI 1 and AEI. Goat--normal nonimmune IgG and anti-CMV.

ABBREVtATIONS: AN, anmiotic nuclear staining; C, cytoplasmic staining; CC, chorionic cell cytoplasnfic staining; HISV, hyperimmunization with solubilized virions; Macro, adrenal adventitial macrophages; N, nuclear staining; ND, not done due to scarcity of target tissue; NSB, nouspecific background staining; Post, postinfection; Pre, preinfection; PVM, perivascular mesenchymal cells; SAC, subamnionic chorion cells; SAM, subamnionic meseuchymal cells; VH, chorionic villias Hofbauer cells; VSM, vascular smooth muscle cells; Neut Tit, neutralization titer.

* Lot of anti-HSV 2 IgG used to screen all patient tissues. Staining grade of 1 to 4: Positive cell number and intensity of stain were compared with the staining observed in tissues stained by the lot 1188 rabbit anti-HSV 2 IgG as follows: 4 +, 75-100%; 3 +, 50-75%; 2 +, 25-50%; I +, less than 25%.

Endothelial nuclei were positive in three of 380 abortion cases ( fig. 6 ); the rabbit anti-HSV 2 antibody stained specimens in all three cases, while staining with the mouse anti-ICP4 antibody was observed in only two cases. Weak intranuclear staining of amnionic cells in the placental target cells was observed with the mouse anti-ICP5 antibody, which detects nuclear antigen in HSV 1 and 2 acute infection in cell culture.

5. The Dako HSV 2 individual sera were from rabbits that had not had the extensive repeat immunizations undergone by tile rabbits in the pool of the screening antibody (lot 1 188). They were, therefore, of much lower titer. The "new" lot (no. 012C) of Adsorption experimer~ts. Two identical adsorption experiments were periormed four months apart with coded samples of noninfected frozen cells, or cells infected with CMV or HSV l or HSV 2. The thawed cell lysates were mixed with the screening HSV 1 and HSV 2 antibodies for 30 minutes at room temperature and centrifuged, and the supernatant fluid was tested on the target tissues. In the first experiment, the anti-HSV 2 antibody was easily identified by decreased staining, compared with that produced by the nonadsorbed antibody assayed at the same time in serial sections of target tissues. In the second experiment, no reduction in any of the adsorbed antibodies was detected. The reason for the discrepancy is not known, although adsorption experiments such as this can be difficult to replicate and interpret, because of the frozen, then thawed, nonfixed antigens and complement receptors.

False-positive res'tdts. Only two false-positive staining patterns were identified, both involving the endometrial epithelium. The first, and more frequent (20 of 380 cases), was a basal cytoplasmic granular staining pattern, without the usual diffuse and/ or apical staining ( fig. 10 ). This pattern occurred with several of the HSV-specific and nonspecific anti-bodies and was independent of the secondary antibody, ABC complex, and colorizing solution. The mechanism of staining is not known. The second, and quite rare (two of 380 cases), false-positive pattern was confined to endometrial nuclei in a focal pattern ( fig. ] 0) . This intranuclear staining was independent of the primary and secondary antibodies, the enzymes attached to the ABC complex (peroxidase, alkaline phosphatase, and glucose oxidase), and the various colorizing solutions. It was dependent only on the ABC complex itself. The mechanism remains unknown, because neither biotin or avidin preincubations blocked the staining. No viral cytologic abnormalities were detected in any of the 200 specimens. Material from spontaneous abortiops was more inflamed than that from therapeutic abortions (92 versus 66 per cent, P < 0.001), but there was no difference between the two groups with regard to acute versus chronic choriodeciduitis (P "-0.565). Although total HSV positivity (endometrial epithelial positivity and/or placental subamnionic chorion positivity and/or chorionic villus positivity) was similar in the material from spontaneous and therapeutic abortions (48 and 41 per cent, P = 0.386), placental subamnionic chorion positivity and chorionic villus positivity were significantly more frequent in spontaneous abortion specimens (39 and 14 per cent) than in those from therapeutic abortions (13 and 0.0 per cent, P < 0.001 and P < 0.001, respectively), while endometrial epithelial positivity was more frequent in therapeutic than in spontaneous abortion material ( Table 3 lists the statistical correlations between the endometrial phases, abnormalities, and inflammation for endometrial epithelial negativity and endometrial epithelial positivity for 180 cases. There was no significant difference between the ages of the two groups (HSV-negative, 41.5 years; HSV-positive, 39.8 years; P = 0.765) or for the volume of endometrial tissue removed, excluding blood clots and mucus (<3 mm 3 versus 3 to 5 mm 3 versus >5 mm 3, P = 0.797).

The detection of endometrial epithelial positivity was significantly different in tile different phases of the cycle. Normal proliferative (I0 per cent), hyperplastic (14 per cent, P = 0.697), and atrophic (6.7 per cent, P > 0.990) endometria were not statistically different. Normal secretory (40 per cent) and abnormal secretory (67 per cent) endometria were significantly different from normal proliferative (P = 0.017 and <0.001), hyperplastic (P = 0.001 and <0.001), and atr-ophic (P = 0.014 and <0.001) endometria, as well as from each other (P = 0.040). Furthermore, the cycle date in the cases of normal secretory endometrium was significantly earlier for the HSV-negative cases (day 20.6) than for the HSV-positive cases (day 22.6, P = 0.004), suggesting activation of HSV protein synthesis in the latter portion of the normal secretory phase. The percentage of HSV positivity in normal secretory phase endometriuln (40 per cent) was similar to that found in material from both therapeutic (41 per cent) and spontaneous (48 per cent) abortions. The presence of HSV positivity in chronic active endometritis depended on the phase and was similar to the degree of HSV positivity found in the corresponding noninflamed normal phases. * PARAMETERS: Total inflammation, presence of acute or chronic choriodeciduitis (CD); normal, no acute or chronic CD; total I ISV +, staining in one or more of the following patterns--chorionic villus llofbauer cells (VH), snbamnionic chnrion (SAC), or endometrial epithelium (EMET) (see Results section anti figures I and 3); blighted ovmn, destruction of the embr)o as defined by lack of nucleated embryonic erythrocytes and capillaries in the chorionic villi; muhiparity, more than one pregnancy; SAb and TAb, spontaneous and therapeutic abortions, respectively; recurrent SAb, more than one spontaneous abc)rtion; gestational age, the gestation:d age of the embryo at the time of the abortion; patient or partner cold sore, clinical history of at least one episode of ~wal HSV infection in the patient or partner at the time of pregnancy, respectively; patient or partner genital tlSV, clinical Ifistory of genital HSV infection in the patient or partner, respectively.

i The patient ages were normally distributed for all categories and were analyzed t))" the Anderson-Darling test, while all other parameters were nonparametric and were analyzed by the Wilcoxon mmpaired rank sum test.

The specificity of the screening rabbit anti-HSV antibody in detecting only HSV antigens underlies the concepts about latent intrauterine and neonatal HSV infection presented in this study. Our contention that the antigens that we detected in endometrial, placental, and cordal tissues are specific for HSV, probably of type 2 origin, is strongly supported by the control data. The control studies included numerous non-HSV virtts-infected tissues, HSV-negative tissues, and bacterium-infected tissue. In addition, numerous HSV-positive and -negative antibodies derived from different animals and from different routes of infection with HSV viruses of both'type 1 and 2 were used as controls. Confirmation of this HSV specificity will have to be accomplished by DNA and/or RNA extraction and/or in situ hybridization techniques, if the necessary technical sensitivity can be achieved, lt.ta.t5

Mouse monoclonal antibodies lmve been useful in defining the HSV-specific nature of tile detected antigens, especially the amnionic intranuclear staining of ttte immediate/early antigen, ICI'4. Exact delineation of the HSV-specific proteins present in our tissues was difficult, for at least two reasons. First, tltere are 15 to 35 different virion proteins in the HSV virion, 16,t7 and the ntunber of available monoclonal antibodies to these proteins is limited. Virus-specific nonvirion and celhtlar proteins also are present in HSV-infected cells; some of these proteins may have been detected with the Dako antibody, which recognizes both HSV-specific virion and nonvirion antigens, ts The antibody response to HSV is polyclonal, and, in humans, at least 31 HSV 1 and 27 HSV 2 virus-specific proteins can be antigenic with great individual variation, t2

Second, little is known about the fate or degradation sequence of the various HSV-spccific virion and nonvirion proteins in phagocytic cells (primarily macrophages) tltat organize (i.e., clean up) the cellular and viral debris after acute HSV infection. HSV proteins are degraded differentially in different cell lines after acute infections in cell culture. 19 Therefore, redundant antigenic sites on the virus-specific proteins 9°-23 may be either exposed or lost during the degradation process in macroplmges or other phagocytic cells, such as some choriodecidual cells and the Hofbauer cells of the chorionic villi. Tiffs uncertainty may make identification of specific HSV protdins by imnumohistochemistry and/or isolation difficult in subacute or chronic infections in animal tissues. Antigenic variants of the viral glycoproteins may also occur. 24 The long-term (weeks to months) fate of the numerous HSV virion and nonvirion protein antigenicities in acutely infected tissues has not been studied in animal models.

Our working hypothesis is that a latent endometrial epithelial infection is established by HSV type 2 intrauterine infection through one or more of the following routes: ascending cervical, transneural, maternal primary viremia during pregnancy, postnatal primary viremia in childhood or adulthood, and activated endometrial intrauterine embryonic or fetal infection with or without germ cell involvement. Once this latent endometrial infection is established, sporadic transient focal acute infections may be activated, as in the oral and genital regions.

This endometrial activation is maximal at the end of the secretory phase, when endometrial prostaglandin synthesis is maximal 25 and when local immunosuppression is occurring to protect the implanting embryo (L. Olding, unpublished data) if fertilization has occurred. Prostaglandins produce local and systemic immunosuppression, 26 and both immtmosuppression 27 and prostaglandins 8'29 activate latent HSV infection.

The level of endometrial prostaglandin synthesis is also quite high during pregnancy, s° and genital HSV activation and asymptomatic shedding of HSV are increased during the gestational period. 31 Furthermore, the presence of ICP4, the immediate/early protein detected in our study, is necessary, but not sufficient, for HSV activation and replication. Further events are required for virus production, s2,s3 possibly prostaglandin synthesis and/or local immunosuppression. The immune response to HSV infection in both tile pregnant and nonpregnant states is quite complex and involves antibody and complement, antibody and natural killer ceils, macrophages, and polymorphonuclear leukocytesJ '~7 Maternal antibody in the anmionic fluid may protect the embryo or fetus. 34 The cellular immtme response to HSV 2 infection is decreased during pregnancy. 35 If conception has not occurred, the acute activated infection is not detected clinically, because tile endometrium has no sensory pain nerves, and there is no transcervical discharge or spotting. If conception has occurred, however, the infection has the potential to spread to the placenta and amniochorionic sac. If the infection is not inhibited by maternal antibody in the amnionic fluid, 34 it may continue into the embryonic or fetal circulation and/or tissues, as evi-denced by antigen psotivity in the chorionic villi. If the embryo or fetus is infected, the probability of spontaneous abortion, stillbirth, or neonatal complications is high. If the infection is limited to ttle amniochorionic sac, however, tile chance of embryonic or fetal damage is greatly reduced (see companion paper4). Spontaneous abortion and infants small for the gestational age are associated with nonspecific villitis, ~° a condition that we found to be associated with HSV antigen positivity. Specimens from maternal placental floor infarction ~7'3s were HSV antigen-positive in five of the eight cases assayed. Our companion paper 4 describes the neonatal complications associated with HSV antigen positivity.

The observational and clinical data in support of this hypothesis are as follows. First, endometrial epithelial positivity is maximal in tile late secretory phase (day 22.4, 40 per cent positive) as compared with the proliferative phase (10 per cent, P = 0.017), endometrial hyperplasia (14 per cent, P = 0.001), and tile atrophic state (7 per cent, P = 0.014). Maximal positivity occurred in abnormal secretory phase endometria (68 per cent, P = 0.040 versus normal secretory phase), probably a time of altered prostaglandin synthesis. 39 Second, the mean gestational age increases progressively for endometrial epithelial positivity,.placental subamnionic chorion positivity, and chononic villus positivity (8.1 weeks, P = 0.006, versus 9.8 weeks, P = 0.006, versus 11 weeks, P = 0.028, respectively). The increasing gestational age for antigen positivity at these sites supports tile concept that the infection begins in the endometrium and progresses into the anmiochorion and, finally, into tile chorionic villi (i.e., the embryo or fetus).

Third, placental subamnionic chorion positivity is dependent on endometrial epithelial positivity (P = 0.004), and chorionic villus positivity is dependent on placental subamnionic chorion positivity (P = 0.001). This finding supports the concept that amniochorionic infection cannot occur without endometrial infection, and chorionic villus infection cannot occur without amniochorionic infection. The reason for which chorionic villus positivity is not statistically dependent on endometrial epithelial positivity could be that most epithelium-positive endometria may not produce placental subamnionic chorion-positive or chorionic villus-positive infections. Another reason for the discordance in individual specimens between endometrial epithelial positivity, p.lacental subamnionic chorion positivity, and chorlonic villus positivity is that very little of the endometrium in the uterus accompanies the placental tissue at the time of abortion. Furthermore, little of the removed endometrium may be submitted for examination, because of the difficulty in separating it from the placental and decidual tissue during gross examination.

Fourth, material from spontaneous abortions is more likely to be inflamed (92 per cent, P < 0.001) than is that from therapeutic abortions (66 per cent).

Both placental subamnionic chorion positivity and chorionic villus positivity show trends (P = 0.139 and 0.078, respectively) toward an association with inflammation, while endometrial epithelial positivity is significantly associated with noninflamed normal tissue (P = 0.022). Some therapeutic abortions are associated with inflammation and blighted ova, because (hey are incipient spontaneous abortions that have not been recognized clinically. Therapeutic abortions occur earlier than spontaneous abortions (at'8.5 versus 9.2 weeks of gestation, P = 0.038).

Fifth, placental subamnionic chorion positivity (39 versus 13 per cent, P < 0.001) and chorionic villus positivity (14 versus 0.0 per cent, P < 0.001) are more frequent in spontaneous than in therapeutic abortions, while the incidences of endometrial epithelial positivity are similar (26 versus 33 per cent, P = 0.230).

Sixth, blighted ova are associated more frequently with spontaneous than with therapeutic abortions (78 vs 27 per cent, P = 0.001), with acute (48 per cent, P = 0.043) and chronic (73 per cent, P = 0.007) inflammation more frequently than with no inflammation, with placental subamnionic chorion positivity more frequently than with subamnionic chorion negativity (67 versus 44 per cent, P = 0.008), and with chorionic villus positivity more frequently than with chorionic villus negativity (70 versus 49 per cent, P = 0.207, trend), but not with endometrial epithelial positivity or negativity (51 versus 51 per cent, P = 0.820).

The data presented in the fourth, fifth, and sixth points support the concept that HSV infection of the amniochorion (placental subamnionic chorion positivity) and chorionic villi (chorionic villus positivity) can destroy the embryo (blighted ovum) and produce spontaneous abortion, as suggested previously. 4°-43 It is difficult to determine the percentage of spontaneous abortions caused by HSV infection. The identified causes of spontaneous abortion are chromosomal abnormalities (about 50 per cent), bacterial infection (about 10 per cent), CMV (rare), and chlamydial infection (unkown frequency); the cause is unknown in up to 40 per cent of cases)J Since our data show an increase of at least 50 per cent in blighted ova when spontaneous abortion material has placental subamnionic chorion positivity (67 per cent/44 per cent = 152 per cent) and since HSV infection itself may produce chromosomal damage (double-stranded DNA virus), a significant proportion of the 40 per cent of spontaneous abortions of unknown etiology may be caused by HSV infection.

Recurrent spontaneous abortions may be associated with endometrial epithelial positivity (30 versus 28 per cent, P = 0.143) and chorionic villus positivity (50 versus 23 per cent, P = 0.207), but the present data show only a trend, and the number of mothers with recurrent spontaneous abortions in this study was small (38 cases). Recurrent spontaneous abortions are not associated with placental subamnionic chorion positivity (26 vs 22 per cent, P = 0.521). This finding might correlate with the amount of maternal anti-HSV antibody in the amnionic fluid, a4 That is, the infection is blocked at the amniochorionic level, without transmission into the embryo or fetus (i.e., chorionic villi). Recurrent neonatal HSV infection has not been documented in successive pregnancies of the same mother. 43 Two of the HSV-positive patients in this study who had spontaneous abortions subsequently had normal full-term infants with HSV antigen-positive placentae and cords, but with negative chorionic villi. In a guinea pig model of placental CMV infection, only 27 per cent of the fetuses that had CMV-positive placentae were CMV-positive by direct culture. 44 Muhiparity is associated with chronic inflammation more than with the absence of inflammation (79 versus 58 per cent, P = 0.034), and shows a trend toward associations with endometrial epithelial positivity (75 versus 69 per cent, P = 0.179) and placental subamnionic chorion positivity (78 versus 65 per cent, P = 0.084). The greater the number of pregnancies, the greater is the potential for latent endometrial infection (endometrial epithelial positivity) to activate and produce acute infection with intrauterine infection (placental subamnionic chorion positivity). Muhiparity was not associated with increased risk of embryonic infection (chorionic villus positivity, 75 versus 67 per cent, P = 0.635).

The only significant association between oral and genital HSV infections in the patient and/or the father of the aborted embryo or fetus was between genital HSV infection in both individuals (50 per cent of patients with clinical histories of genital HSV infection had partners with clinical histories of genital HSV infection, and 6.7 per cent of patients with positive clinical histories had partners with negative clinical histories, P = 0.001). Only 3.6 per cent of 83 HSV antigen-positive patients and 5.6 per cent of 107 HSV antigen-negative patients had clinical histories of genital herpes infection (P = 0.522). Asymptomatic genital HSV 2 infection, however, is common, occurring in more than 50 per cent of nonpregnant women with primary infections, 45 25 per cent of nonpregnant women with recurrent infections, 46 and 13 to 33 per cent of women during pregnancy, a~,41,48 The optimal time to detect asymptomatic shedding from an activated endometrial infection in nonpregnant women would be the first day of menstrual flow, as maximal antigen detection occurs in the late secretory phase (day 22 to 23).

The data reported in this study suggest that latent intrauterine HSV infection is etiologically important in spontaneous abortion. The accompanying paper describes the neonatal problems produced by infections of this type. 4 Future studies should include the following investigations: determination of maternal anti-HSV antibody status; extraction of DNA from, and in situ DNA/RNA hybridization of, aborted and curettage tissues to determine whether HSV-specific nucleic acid sequences are present; and direct culture of first-day menstrual flow material for infectious HSV.

Note added in proof. In collaboration with Dr. David Myerson, we have detected HSV-specific DNA in amnionic and chorionic nuclei in serial sections from the specimen in figures 1 and 9. In situ hybridization using a biotinylated HSV DNA probe as described in reference 1 1 was used. Evaluation of endo'rnetrial tissues is underway.

+ SAC* (48) 16.7 + TAb (99) 13.1 + EMET + (58) 39.7 + Tot Inflam ÷ (157) 25.6 + Acute CD (98) 23.5 + TAb (97) 8.5 ±

SAC ÷ (44)9.8 ±

+ Chronic CD (45) 73.3 + SAC* (48)66.7 + Partner genital + (6) 50.0 + Chronic CD (56) 78.6 + Prolif (19) 10.5 + Abn Secretory (21) 66.7 + Atrophic (15) 6.7 + Hyperplastic (35) 14.3 + Prolif (19) 10.5 + Hyperplasia (35) 14

+ Chronic CD (58) 29.3 + Tot Inflam* (158) 28.5 + Chronic CD (59) 30.5 + Partner oral + (17) 17.6 + Tot inflam'-(123) 8.1 + Acute CD* (78) 7.7 + Chronic CD ÷ (45) 8.9 + TAb (65) 60.0 + TAb (99) 41.4 + TAb (99) 33.3 + Chronic CD (58) 29.3 + Partner oral* (18) 22.2 + Partner genital ÷ (6) 33.3 Partner oral/genital + (24) 25.0 4 Chronic CD (59) 42.4 + Acute CD (99) 27.0 + Chronic CD (59) 30.5 +

Chronic CD (5-t) 9.2 ± 2

Acute CD (92) 8.6

Chronic CD (54) 9.2 ± 2.7 Blighted ovum + (73) 9.4

Herpes simplex

Neonatal herpes simplex virus infection in King County

RobbJA: A new enzyme immunolectin tissue stain

Intrauterine latent herpes simplex virus infection. II. Latent neonatal infection

Herpetic inclusions in the endometrium

Herpesvirus hominis endometritis in a young woman wearing an intrauterine contraceptive device

Optically clear nuclei, an alteration of endometrial epithelium in the presence of trophoblast

Leukocytoclastic vasculitis assnciated with cutaneous infection by herpesvirus

Kefalides NA: Herpes simplex virus type 1 infection of endothelium reduces collagen and fibronectin synthesis

Cytomegalovirus antigen within human arterial smooth muscle cells

Widespread presence of histologically occult cytomegalovirus

Antibody response in humans to individual proteins of herpes simplex viruses

Pathogenic murine coronaviruses III. Biological and biochemical characterization of temperature-sensitive mutants of JHMV

Detection of herpes simplex virus in clinical specimens by DNA hybridization

Localization of herpes sireplex virus in tile trigeminal and olfactory systems of tile mouse central nervous system during acute and latent infections by in situ hybridization

Herpesviruses and their replication

Dako Labs rabbit anti-HSV 1 and 2 IgG package insert. November 1975 (5 references listed in insert from the Scandinavian literature

Monoclonal antibodies to herpes simplex virus type 1 proteins, including tile immediate-earl)' protein ICP4

Serological analysis of herpes simplex virus types 1 and 2 with monoclonal antibodies

Localization of a type-specific antigenic site on herpes simplex virus type 2 glycoprotein D

Extensive homology between the herpes simplex virus type 2 glycoprotein F gene and the herpes simplex virus type 1 glycoprotein C gene

Antibodies to a synthetic oligopeptide that react with herpes simplex virus type 1 and 2 glycoprotein C

Antigenic variants of herpes simplex virus selected with glycoprotein specific monoclonal antibodies

Best FA, et ah Prostaglandins in endometrium and menstrual fluid from normal and dysmenorrhoeic subjects

Regulation of the immune response by prostaglandins, a review

Herpes simplex virus. In Fields BN

Reactivation of herpes simplex virus infection by uhraviolet light and possible involvement of prostaglandins

The effect of prostaglandins on the multiplication and cell-to-celt spread of herpes simplex virus type 2 in vitro

Prostaglandins and the regulation of uterine blood flow in pregnancy

Genital herpes in pregnancy: risk factors associated with recurrences and asymptomatic viral shedding

Expression of herpes simplex virus type 2 antigens in premalignant and malignant human vulvar cells

Cells that constitutively express the herpes simplex virus intmediate-early protein ICP4 allow efficient activation of viral delayed-early genes in traits

Neutralization of herpes simplex virus by antibody in amniotic fluid

Humoral and cell-mediated responses to herpesvirus antigens during pregnancy--a longitudinal stud

Clinical and pathologic aspects of recurrent placental villitis

Recurrent maternal floor infarction: a preventable cause of fetal death

Maternal floor infarction. HUM PATHOt

The menstrual stimulant in puberty

Mfiternal herpes-simplex infection causing abortion: INTRAUTERINE HSV INFECTION: SPONTANEOUS ABORtiON (Robb et ol.) histopathologic study of the placenta

Perinatal risk associated with maternal genital herpes simplex virus iqfection

Pathology of the Female Genital Tract

Herpes simplex

The placenta as a site of cytomegalovirus infection in guinea pigs

Neonatal herpes simplex virus infection of the central nervous system

Genital herpes simplex virus infections: clinical manifestations, course, and complications

Herpesvirus hominis type II infections in asymptomatic pregnant women

Herpesvirns infections of pregnancy

Enzyme immunofihration technique for rapid diagnosis of herpes simplex virus eye infections in a rabbit model

Complement-requiring neutralizing antibody in guinea pigs with primary and recurrent genital herpes

Neonatal herpes simplex virus infection in guinea pigs