336

(49 per cent). Red areas on the cervix were seen before iodine staining in both groups.

Pathology. Vaginal adenosis was detected by biopsy in 35 per cent of the exposed but only 1 per cent of the control subiects In the former the adenosis was usually multifocal. On microscopical examination it was characterized by the presence of a single layer of mucinous columnar cells of endocervical type or a single or pseudostratified layer of cells that resembled endometrial and tubal epithelial cells. The latter cells were often ciliated; most of them had dark, deeply eosinophilic cytoplasm, but some had clear cytoplasm. In occasional cases, minor degrees of atypicality or loss of polarity of the nuclei were observed. The columnar epithelium lined the surface of the vagina or formed glands in the lamina propria (Fig. 1). Often, the glands were largely or totally replaced by solid masses of squamous epithelium (squamous pegs), which were continuous with the surface squamous epithelium (Fig. 1); the epithelium both in the pegs and on the surface was almost always composed of glycogen-poor or glycogen-free basal and prickle cells. This characteristic of the squamous cells or in some cases the replacement of the surface squamous epithelium by a layer of glandular epithelium accounted for the lack of staining with iodine. Two histologic variants of adenosis that were observed in the squamous pegs or in the surface squamous epithelium were droplets of mucin within individual squamous cells

Figure 1. Vaginal Adenosis (Hematoxylin and Eosin Stain). Glands with ciliated dark cells that resemble the endometrial or tubal lining lie in the Jamina propria and merge in areas with squamous pegs (arrow). Both the squamous epithelium that lines the surface of the vagina and the squamous pegs continuous with it are composed of glycogen-free basal and prickle cells.

Feb. 13, 1975

(Fig. 2) and small pools of mucin surrounded by tall to inconspicuously flat cells with mucinous cytoplasm (Fig. 3). The mucin droplets and pools, which sometimes became apparent only after mucin staining, were present in 74 per cent of the subjects with adenosis and were the only evidence of it in 29 per cent of these cases (8 per cent of all the exposed subjects). Squamous pegs were observed in 76 per cent of the exposed subjects with biopsy-proved adenosis. Although pegs were observed in the vaginal biopsies of 16 per cent of the exposed subjects without adenosis, they were not observed in the unexposed.

Adenosis was detected in only one control, and it differed from the typical form of the disorder in the exposed subjects. In that case a single cystic gland lined by mucinous epithelium lay in the lamina propria; the overlying squamous epithelium was normally glycogenated, and no squamous pegs were seen.

In contrast to vaginal adenosis, the cervical erosions were morphologically similar in the exposed and the control subiects, being characterized by an epithelium that was almost always mucinous; cells of tubal or endometrial type were only occasionally seen. The glandular epithelium lay on the surface, sometimes covering micropapillae, or lined glands in the lamina propria. There were however. quantitative differences in the erosions of the two groups of subiects Squamous pegs were more frequent in the cervical biopsies of the exposed (43 per cent) than in the control series (6 per cent). Also, mucinous droplets and pools were more commonly encountered in the for

mer.

Cytology. Quantitative differences also existed in the cytologic specimens of the two groups. In the vaginal scraping, columnar cells, metaplastic squamous cells, squamous cells containing mucous droplets, or dysplastic cells were found in 9 per cent of the exposed subjects (11 per cent in those with adenosis and 7 per cent in those without), in contrast to 1 per cent of the controls. In the aspirates of the vaginal pool, these abnormalities were also observed more frequently in the exposed (19 per cent) than in the control subjects (9 per cent), being present

Figure 3. Vaginal Adenosis (Mucicarmine Stain). One pool of mucin is surrounded by flat cells with mucinous cytoplasm that were obvious only with mucin stains (upper left), whereas a second pool is surrounded by cells with foamy cytoplasm obvious both with hematoxylin and eosin and with mucicarmine stains (lower right).

in 27 per cent of the exposed girls with biopsy-proved adenosis. Although mild squamous-cell dysplasia was observed in the vaginal scrapings from three exposed subjects and in the vaginal aspirate from a fourth, this change was not confirmed in any by multiple punch biopsies.

Other factors. The increased occurrence of adenosis, erosion, non-iodine-staining areas, and ridges in the exposed in comparison to the unexposed subjects was investigated further to determine whether other features (e.g., intrauterine x-ray exposure, smoking by parents and daughters, or maternal medications other than diethylstilbestrol) could have also had a role. No association with any factor other than intrauterine diethylstilbestrol could be identified.

Investigation of the Total Exposed Population of 133 Subjects (13 to 25 Years of Age)

Eighteen exposed white female subjects 13 to 17 and five non-white subjects 19 to 24 years of age were also examined, and the findings were combined with those of the 110 exposed subjects from the controlled study to provide the results of the investigation of all the post-menarchal females from the Diethylstilbestrol Clinic who agreed to be examined. Non-iodine-staining areas of the vagina were detected in 80 (60 per cent), and biopsy-proved adenosis in 49 (37 per cent). Non-iodine-staining epithelium was found on the cervix in 126 (95 per cent),

337

and biopsy-proved erosion or ectropion in 112 (84 per cent). Fibrous ridges of the vagina and cervix were found in 28 (21 per cent) of the subjects. No cases of cancer were detected. Only seven (5 per cent) of the subjects had pelvic examinations that were normal on inspection and after staining with iodine solution.

Table 4 examines the relation between the week of initiation of diethylstilbestrol therapy and the proportion of subjects having abnormalities of the vagina or cervix. Vaginal adenosis was more frequent in those daughters whose mothers began diethylstilbestrol in early pregnancy. It was observed in 73 per cent of subjects initially exposed during the first two months of pregnancy, but in only 7 per cent of the 30 initially exposed in the 17th week or later (Table 4). None of 19 subjects whose mothers began therapy in the 18th week or later were found to have adenosis. The linear regression of the proportion of those found to have adenosis with respect to the month of initiation of therapy was highly significant (p <0.001). Similarly, cervical erosion was more frequent when therapy was begun in early pregnancy, but the regression (p < 0.01) was not as striking as for vaginal adenosis (Table 4). This result can be explained in part by the fact that cervical erosion often occurs in the absence of diethylstilbestrol exposure (Table 3). No association was demonstrated between the time of initiation of therapy and the presence of vaginal or cervical fibrous ridges, and one ridge was detected in a subject whose mother began treatment as late as the 21st week. The location and appearance of the ridges, however, varied to some extent with the time therapy was initiated. Nine of the 10 daughters with ridges who were exposed before the 12th week had a concentric ridge surrounding the external os of the cervix, with the formation of a pseudopolyp; the 10th had a transverse narrowing in the upper vagina. Both pseudopolyps and cervical ridges that were only partly circumferential were observed in subjects exposed between the 13th and 16th weeks of gestation. None of the four with ridges whose mothers started treatment in the 17th week of pregnancy or later had a cervical pseudopolyp, but all had partial ridges on the cervix or in the vagina. Because the mothers who were treated in the Diethylstilbestrol Clinic maintained a strict and identical dosage schedule, the total dosage of maternal medication was directly related to the time in pregnancy diethylstilbestrol was begun, so that it was impossible in this study to separate the effect of dosage from that of the onset of medication.

The analysis by the use of a multiple logistic model12 revealed no correlation of vaginal or cervical abnormalities with the age of the subjects at the time of examination or with their sexual histories. As individual factors, the

$338 :

age of the subjects at the time of examination, as well as the week in pregnancy in which diethylstilbestrol was begun, did affect the incidence of adenosis. However, the mothers of the younger subjects began diethylstilbestrol earlier in pregnancy, resulting in a high correlation between subject age and the week of initiation of maternal medication. After adjustment of the incidence of adenosis for the week in which diethylstilbestrol was started, there was no residual effect owing to the age of the subjects. Vaginal adenosis, when adjusted for the week in which the estrogen was started, was detected less frequently in patients who had used oral contraceptives (chi-square 5.05, p < 0.05), but the use of oral contraceptives had no effect on the frequency of non-iodine-staining areas in the vagina. Cervical erosion, adjusted for the week in which diethylstilbestrol was started, was less frequent among those who had been pregnant (chi-square 5.44, p < 0.05). Pregnancy did not affect the incidence of vaginal adenosis or of non-iodine-staining areas in the vagina.

DISCUSSION

=

=

The results of this investigation of diethylstilbestrolexposed and unexposed female subjects are consistent with previous uncontrolled observations that vaginal adenosis as well as transverse vaginal and cervical ridges are very rare in young females in the absence of intrauterine exposure to diethylstilbestrol. 46 Although cervical abnormalities (failure of the portio mucosa to stain with iodine and erosion) were encountered in almost all the exposed subjects in this study, these findings were less specific, being present in about half the controls. There were no significant differences between the exposed and control groups in postnatal medical histories, including the onset and frequency of menstrual bleeding. These limited data suggest that ovarian function is not altered, at least in the first three decades of life, as a result of prenatal exposure to diethylstilbestrol. Similarly, the fact that there was no significant difference in health histories of the mothers in each group up to the present time, including the development of cancer after the ingestion of diethylstilbestrol, suggests that the medication did not affect their health. This interpretation, however, is based only on interview histories and not on physical or laboratory examinations.

The gynecologic and pathological findings of the present study were based on inspection, iodine staining, cytologic sampling and biopsies of the vagina and cervix. In view of the fact that scraping the vagina yielded cells compatible with vaginal adenosis (columnar cells or mucin droplets within squamous cells) in only 11 per cent of the cases of adenosis, and aspiration of the vaginal pool in only 27 per cent, cytology must be considered unreliable as a screening test for this disorder. Nevertheless, the identification of these cell types in a routine vaginal-pool aspiration, as well as the failure of the vagina to stain with iodine during a pelvic examination, should raise a high index of suspicion of diethylstilbestrol exposure in a young woman, who may be unaware of her prenatal history. The additional performance of mucin staining of biopsy specimens of the vagina facilitates the detection of mucin

Feb. 13, 1975

droplets and pools in squamous epithelium, permitting the pathologist to recognize adenosis more easily than with the use of routine hematoxylin and eosin stains alone.

The failure to detect vaginal adenosis in any subject whose mother began therapy during the 18th week of pregnancy or later suggests that this disorder is congenital, reflecting a derangement of vaginal development related to the action of diethylstilbestrol before that time. 13 This observation correlates well with data from the Registry of Clear-Cell Adenocarcinoma of the Genital Tract in Young Females, in which no cancers have vet been uncovered in female subiects exposed to diethylstilbestrol or similar drugs later than the 17th week of gestation.” Although it was not possible from the current study to separate the effects of the total dosage of diethylstilbestrol from the time of its initiation during pregnancy, previous uncontrolled studies4.5 as well as data from the Registry3 have provided evidence that neither adenosis nor carcinoma is dose related, within the wide range of the amounts of diethylstilbestrol that are known to have been prescribed.

The incidence of vaginal adenosis that has been reported in diethylstilbestrol-exposed female subjects has varied. For example, Stafl and his associates found it in 57 of 63 exposed subjects on colposcopic and biopsy examination. Their results are in contrast with those of Townsend, who also used the colposcope but demonstrated adenosis in only 30 per cent of 286 exposed females; he identified a metaplastic squamous transformation zone corresponding to non-iodine-staining areas of the mucosa in an additional 30 per cent (personal communication). These results are similar to those of the present study, in which the method of examination was palpation and iodine staining followed by biopsy of abnormal areas. Some of the reported variations in incidence of adenosis may be related to differing estimates by the examiners of the boundary between the vagina and the cervix, which may be difficult to define in the presence of ridges, as well as differences in histologic terminology, the time of initiation of diethylstilbestrol therapy, and the population examined (e.g., unselected, asymptomatic, exposed subjects vs. symptomatic patients or those referred because of abnormalities noted by other physicians).

A proposed mechanism by which intrauterine exposure to diethylstilbestrol acts to result in vaginal adenosis is inhibition of the normal replacement of the embryonic müllerian epithelium of the vagina by the upgrowing squamous epithelium of the vaginal plate, which may be of either wolffian or urogenital-sinus origin.4.8 Normally, in the unexposed mature female, the transformation zone between the two types of epithelium is at or near the external os of the cervix. It has been suggested that cervical erosion and vaginal adenosis in exposed subjects represent simply an increasingly more distal positioning of the transformation zone onto the cervical portio and vagina, respectively. If this hypothesis were true, one might expect that cervical erosion and vaginal adenosis would be identical on histologic examination, but differences do exist. For example, in the biopsy specimens of cervical erosion in the present investigation, almost all the glan

6

Vol. 292 No. 7

dular epithelial cells had the appearance of mucinous cells of the endocervix, whereas in vaginal adenosis, many of them resembled those of the endometrium or fallopian tube. These findings suggest that the pathogenesis of cervical erosion and vaginal adenosis may not be identical, and that the two processes may differ in biologic behavior on long-term investigation.

The data from the current study indicate a decreased frequency of vaginal adenosis in exposed subjects who have used oral contraceptives. This apparent healing might be related to the progestational component of these agents, in view of the suggestive evidence provided by a recent pilot study that healing of vaginal adenosis is promoted by the use of progesterone-theobroma oil suppositories. However, in spite of these preliminary findings, we have not advised the routine use of these agents because of the uncertainty regarding their long-term effects. The decreased frequency of cervical erosions observed in the exposed subjects in this study who had been pregnant might have been due to cervical cauterization after delivery, rather than to the effects of the pregnancy itself, but the inaccurate histories of that procedure did not allow testing of this hypothesis.

It has been suggested that replacement of glands by squamous epithelium may be a mechanism by which vaginal adenosis and cervical erosion heal,6.7.14.15 and because metaplastic squamous epithelium was observed frequently, it was anticipated that these disorders might be less frequent in older subjects as a consequence of spontaneous healing. Although the data afforded no evidence in support of this expectation, all the subjects in this study were younger than 26 years of age at the time of examination, and follow-up observation as they age further may reveal spontaneous healing.

Because the presence of vaginal adenosis has been demonstrated in almost all the cases of clear-cell adenocarcinoma of the vagina in young women when adequate tissue has been available for examination, and because cervical erosion is a consistent finding in association with clear-cell adenocarcinoma of the cervix,3 it is logical to assume that the glandular tissue characterizing adenosis and erosion may provide the base from which these adenocarcinomas develop. Although glandular atypicality has been reported in the vicinity of one clear-cell adenocarcinoma of the vagina, 16 to our knowledge no case has yet been recorded in which a direct temporal or spatial transition from adenosis to adenocarcinoma has been documented in a diethylstilbestrol-exposed woman. In the present study, only minor degrees of glandular atypicality were observed in the biopsy specimens of the cervix and vagina. The results of many studies continue to show that non-neoplastic cervical and vaginal abnormale ties are common in the exposed population, whereas clear-cell adenocarcinomas and glandular abnormalities that can be recognized morphologically as precancerous.

are very Lare.

An additional hazard to diethylstilbestrol-exposed female subjects has been suggested by some authors17 who have proposed that squamous-cell carcinomas also may occur with increasing frequency as these patients age, in

339

view of the recent rare findings of squamous dysplasia and carcinoma-in-situ of the cervix and vagina in them (Stafl and Mattingly, 17 Lanier et al., 18 Robboy, S.J., Scully, R.E., and Herbst, A.L.: unpublished data) and the knowledge that the peak incidence of such alterations in the general population is at an older age than that of exposed women at present. In this study, only four of the 110 exposed subjects (4 per cent) had squamous dysplasia on cytologic examination, and it was slight in all four. With only sporadic reports of these precancerous or possibly precancerous squamous changes, it is impossible to be certain how often they will either progress to invasive cancer or prove to be only a transitory phenomenon related to active squamous metaplasia.

Obviously, many years may be required to determine the risk of development of genital cancer in diethylstilbestrol-exposed female subjects. During this time, thorough periodic examination is clearly indicated.

We are indebted to Dr. Theodore Colton and Dr. Klim McPherson, of the Department of Preventive and Social Medicine, Harvard Medical School, for assistance with statistical analysis, and to Nancy Dreyer Arkin and Jean Sheridan for their efforts in the completion of this study.

REFERENCES

1. Herbst AL, Ulfelder H, Poskanzer DC: Adenocarcinoma of the vagina: association of maternal stilbestrol therapy with tumor appearance in young women. N Engl J Med 284:878-881, 1971

2. Greenwald P, Barlow JJ, Nasca PC, et al: Vaginal cancer after maternal treatment with synthetic estrogens. N Engl J Med 285:390-392, 1971

3. Herbst AL, Robboy SJ, Scully RE, et al: Clear-cell adenocarcinoma of the vagina and cervix in girls: an analysis of 170 Registry cases. Am J Obstet Gynecol 19:713-724, 1974

4. Herbst AL, Kurman RJ, Scully RE: Vaginal and cervical abnormalities after exposure to stilbestrol in utero. Obstet Gynecol 40:287-298, 1972

5. Pomerance W: Post-stilbestrol secondary syndrome. Obstet Gynecol 42:12-18, 1973

6. Stafl A, Mattingly RF, Foley DV, et al: Clinical diagnosis of vaginal adenosis. Obstet Gynecol 43:118-128, 1974

7. Herbst AL. Robboy SJ, Mac Donald GJ, et al: The effects of local progesterone on stilbestrol-associated vaginal adenosis. Am J Obstet Gynecol 118:607-615, 1974

8. Scully RE, Robboy SJ, Herbst AL: Vaginal and cervical abnormalities including clear-cell adenocarcinoma, related to prenatal exposure to stilbestrol. Ann Clin Lab Sci 4:222-233, 1974

9. Smith OW, Smith G van S, Hurwitz D: Increased excretion of pregnanediol in pregnancy from diethylstilbestrol with special reference to the prevention of late pregnancy accidents. Am J Obstet Gynecol 51:411-415, 1946

10. Smith OW: Diethylstilbestrol in the prevention and treatment of complications of pregnancy. Am J Obstet Gynecol 56:821-834, 1948

11. Snedecor GW, Cochran WG: Statistical Methods. Sixth edition. Ames, Iowa State University Press, 1967, pp 246-247

12. Walker SH, Duncan DB: Estimation of the probability of an event as a function of several independent variables. Biometrika 54:167-179, 1967

13. Kurman RJ, Scully RE: The incidence and histogenesis of vaginal adenosis: an autopsy study. Hum Pathol 5:265-276, 1974

14. Burghardt E: Early Histological Diagnosis of Cervical Cancer. Philadelphia, WB Saunders Company, 1973, pp 8, 61-65

15. Coppleson M. Pixley E, Reid B: Colposcopy: A scientific and practical approach to the cervix in health and disease. Springfield, Illinois, Charles C Thomas, 1971

16. Barber HRK, Sommers SC: Vaginal adenosis, dysplasia, and clearcell adenocarcinoma after diethylstilbestrol treatment in pregnancy. Obstet Gynecol 43:645-652, 1974

17. Staff A, Mattingly RF: Vaginal adenosis - a precancerous lesion? Am J Obstet Gynecol 120:666-677, 1974

18. Lanier AP, Noller KL, Decker DG, et al: Cancer and stilbestrol: a follow-up of 1719 persons exposed to estrogen in utero and born in 1943-59. Mayo Clin Proc 48:793-799, 1973

Lucile Kirtland Kuchera, MD

T

his communication presents the follow-up findings in 1,000 women of child bearing age who were given, within 72 hours of sexual exposure, 25 mg of diethylstilbestrol twice daily for five days. Tables are presented giving the time of exposure in relation to the menstrual cycle (Table 1), prophylaxis, if any (Table 2), side reactions to the medication (Table 3), patients using this therapy more than once (Table 4), and the character of the following menses (Table 5). It was explained to the patient that, though this was not a new drug, it was a new use and a follow-up was expected. Possible undesirable effects were discussed before it was prescribed. It was given only to consenting women. It was emphasized to the patient that this was an emergency type of treatment, and, should she have a continuing need for contraception, she should use other

means.

To evaluate its effectiveness, one needs to know the incidence of pregnancy from one act of unprotected sexual intercourse in the women of child bearing age. From analyzing available statistics, C. Tietze, MD, ar

From the University of Mien ran Health Ser

Reprint requests to 207 Fietcher Ave, Ann Ar bor, Mich 4104 (Dr. Kuchera).

562 JAMA, Oct 25. 1971 Vol 218. No 4

rived at the probability of conception from a single unprotected coitus to be between 1 in 50 and 1 in 25.'

The problem of requests for postcoital help by the female patient when she has been exposed to unprotected sexual intercourse for whatever reasons-rape, neglect to use prophylaxis, or failure of prophylaxis, such as condom breakage is familiar and challenging to most practicing physicians. Formerly in this clinic we had only told the patient to wait for her next menses-not very consoling advice to a woman who does not want to be pregnant.

Great credit is due Morris and the New Haven group for their work on effective postcoital antifertility agents. The antizygotic action of es

Table 1.-Time of Exposure in Relation to the Menstrual Cycle

trogens on various animal species has been known for many years."*

My study of patients using diethylstilbestrol for postcoital contraception at our Health Service began in the fall of 1967. In each of the 1,000 cases, an interview or letter was obtained following the ingestion of diethylstilbestrol, so that it was deiinitely known that her next menses came and that she was not pregnant.

The therapy was under the pian and direction of our consulting gynecologist from the University of Michigan Medical Center. The therapy and follow-up was continued by our staff and the data were collected and analyzed by the author.

In the tables when the data are