R. Don Gambrell, Jr, MD, Robert C. Maier, MD, and Barbara I. Sanders, RN.
From the Department of Endocrinology, Medical College of Georgia,Augusta, Georgia; and the Department of Obstetrics and Gynecology,Wilford Hall USAF Medical Center, Lackland AFB, Texas.
The opinions and assertions contained herein are those of the authorsand are not to be construed as official or reflecting the views of the UnitedStates Air Force or the Department of Defense.
Address reprint requests to: R, Don Gambrell, Jr, MDDepartment of EndocrinologyMedical College of GeorgiaAugusta, GA 30912.
Submitted for publication Jamian/ 3, 2983. Revised June 27, 1983. Accepted for publication July 3, 1983. Copyright © 1983 by The American College of Obstetricians andGynecologists.
Abstract
In a prospective study at Wilford Hall USAF Medical Centerfrom 1975 to 1981, 5563 postmenopausal women were followed for a total of 37,236 patient-years of observation.During these seven years, 53 patients were found to havebreast cancer, for an incidence of 142.3:100,000 women peryear. The mean age (± SD) of the patients with cancer was56.9 ± 8.24 years, and the mean age of the entire patientpopulation was 56.8 ± 6.75 years. The expected incidence ofbreast cancer in this age group, according to the ThirdNational Cancer Survey (1975), is 188.3:100,000 women, andfor ages 55 to 59, according to the National Cancer InstituteSurveillance, Epidemiology, and End-Result Reporting(NCI SEER) data (1980), is 229.2:100,000. The lowest incidence of breast cancer (67.3:100,000) was observed in theestrogen-progestogen users and was significantly lower than that of the untreated group (342.3:100,000), with P ≤ .01. The incidence of the estrogen-progestogen users wasalso significantly lower than that expected from the NCISEER data, with a relative risk of 0.3 (95% confidenceinterval, 0.1 to 0.8). The incidence of mammary malignancyin the estrogen users (141.0:100,000) was significantly lowerthan in the untreated group (342.3:100,000), with P ≤ .01.Although the incidence in the estrogen users was not significantly lower than that expected according to the NCI SEER data (relative risk = 0.7, 0.5 to 1.1), there was a trend in thatdirection. These data indicate that estrogen therapy forpostmenopausal women does not increase the risk of breastcancer and may afford some protection. Added progestogento postmenopausal estrogen therapy significantly decreasesthe risk for this malignancy. (Obstet Gynecol 62:435, 1983)
Introduction
The incidence of breast cancer increases throughoutthe female life span and will strike one in 11 women inthe United States.[1,2] In this respect, the incidence of this malignancy differs from cervical, endometrial, andovarian cancer, which peak in the 40s, 50s, and 60s,respectively, and then either decline or plateau. Thesharpest increase in the frequency of breast canceroccurs between ages 30 and 50, or about the time ofmenopause. Then the incidence continues to risesteeply throughout the postmenopausal years, whenestrogen levels are lower. Breast cancer is not only themost frequent malignancy in females (27% of all femalecancers) but also the leading cause of death fromcancer in women (19% of all female cancer deaths) inthe United States.[3] The American Cancer Society estimated that 115,000 new cases of breast carcinomawould be diagnosed in the United States during 1983,and that 38,000 women would die from this tumor inthat year.[3]
Estrogens influence the growth of normal breasttissue. The presence of estrogen receptors and, morerecently, progesterone receptors in breast cancerserves as a prognostic indicator.[4] The presence ofestrogen receptors is related to a longer disease-freeinterval and decreased mortality. It is paradoxical thatsome patients with metastatic carcinoma of the breastrespond to endocrine ablative surgery whereas othersmay have a remission with estrogen therapy. Thepresence of both estrogen and progesterone receptors in breast tumors helps predict a 70% chance offavorable response to endocrine manipulation witheither ablative surgery or antiestrogen therapy. Whenthe estrogen-endometrial cancer controversy was renewed in the mid-1970s, interest was also aroused inthe relationship of estrogen replacement therapy andbreast cancer.[5-7] Long-term studies of large numbers ofwomen have failed to incriminate exogeneous estrogentherapy for any significantly increased risk of breastmalignancy. Unfortunately, none of the epidemiologicstudies have looked at the possible protective effect ofadded progestogens for breast cancer.
Since both the breast and endometrium are targetorgans for the action of hormones, attempts have beenmade for years to define the role of the sex steroids inthe etiology of cancer in these tissues. It is nowaccepted that estrogen replacement therapy increasesthe risk of endometrial cancer.[8-11] Studies from Wilford Hall USAF Medical Center, as well as others, haveshown the efficacy of progestogens in reducing the riskof endometrial cancer in estrogen-treated postmenopausal women.[12-17] With this data base of hormoneusage, patients with breast cancer at this institutionduring the seven years of prospective study have beenreviewed and comprise the basis for this report.
Materials and methods
A prospective study was begun in 1975 to determinethe incidence of breast and endometrial cancer inpostmenopausal women using various hormone regimens. A postmenopausal hormone survey card wasinitiated at the first visit and updated at each subsequent visit. Information obtained included age, parity,blood pressure, weight, and, if previously performed,date of hysterectomy with or without adnexal surgery.Hormone therapy, if any, was recorded, includingtype, dosage, how taken, and for how many years.Age at menopause and age of initial estrogen treatment were listed. A total of 5563 postmenopausalwomen was registered in the study from 1975 through1981.
The patients with breast and endometrial cancerwore identified from the tumor registry. These recordsindude narrative summaries, hospital chart coversheets, surgery reports, pathology reports, tumorboard decisions, radiotherapy records, SouthwesternOncology Group (SWOG) chemotherapy protocols,and copies of all follow-up visits. Once a patient isentered into this registry, outpatient records are codedso that copies of all pertinent records are forwarded lorinclusion. In addition, semiannual or annual questionnaires are mailed to all patients for current status alterthe initial therapy, and patients are seen at leastannually for a minimum of ten years.
Wilford Hall USAF Medical Center is the major AirForce hospital for the San Antonio, Texas area, so thepatient population consists of female personnel onactive duty and wives and other dependents of military personnel, including retired personnel from thatarea. Hormone therapy selected for patients was basedprimarily on symptoms, with one exception. Becauseof the recognized protection from endometrial hyperplasia and neoplasia with cyclic progestogens, thishormone was added to the therapy of more patientseach year as the study progressed. The progestogen challenge test[12] was administered to women withintact uteri, including both those receiving estrogenreplacement therapy and asymptomatic women presenting for annual evaluation. The progestogen administration was continued cyclicly each month aslong as withdrawal bleeding resulted. Patients whohad a previous hysterectomy were usually not treatedwith progestogens. Otherwise, treatment was basedon the patient's symptoms. If the primary symptomwas hot flushes, then oral estrogens were administered. Estrogen vaginal cream was prescribed to patients with atrophic vaginitis, and androgens weregiven to women who had contraindications for estrogen therapy.
Statistical analysis of the data was performed by theSystems and Computer Services, Medical College ofGeorgia, using the test for significance of differencesbetween two proportions and the analysis of variance followed by Tukey's honestly significant difference procedure. Relative risks were estimated bythe method of Mantel[18] and Haenszel. Confidenceintervals were estimated using Miettinen's test-basedmethod.[19]
Results
During the seven years of prospective study from 1975 to 1981, 53 postmenopausal women from the patientpopulation at Wilford Hall USAF Medical Center weregiven a diagnosis of breast cancer. The age of thesepatients, in five-year intervals, is given in Table 1, along with the observed and expected incidence, according to the Third National Cancer Survey (1975)and the National Cancer Institute (NCI) Surveillance,Epidemiology, End Result (SEER) data (1980).[1,2] TheThird National Cancer Survey was conducted beforethe onset of the present survey and the NCI SEER datawere calculated toward the end of the study. The ageof the patients with breast cancer ranged from 31 to 92years, with a mean age of 56.9 ± 1.13 years. All resultsare expressed as mean ± standard error of the mean(SEM). The 31-year-old patient and three of the four jpatients in their 40s had a surgical menopause secondary to bilateral oophorectomy, from two to ten years jbefore the breast cancer was diagnosed. At the time ofthe diagnosis of mammary malignancy, 35 patients jwere currently using hormones (66.0%), and 18 had not received any type of female sex steroids (34.0%).
Table 1. Age of Patient Population With Observed and Expected Incidence of Breast Cancer
The mean age ± SEM for each hormone therapy group, as compared with that of the untreated women, is given in Table 2. Although there were no differences pairwise, there was a statistically significant main effect, with the untreated women (61.3 ± 2.06 years) having a higher age than the other four groups com\bined (P ≤ .02).
Table 2. Mean Age of Patients With Breast Cancer
Of the patients with carcinoma of the breast, 44 wereparous (83.0%), six were nulliparous (11.3%), andthree had conceived but their only pregnancies terminated in abortion (5.7%). Of the 167 pregnancies forthe entire group, 33 (19.8%) ended in abortion, and 44of the 47 patients who had conceived had one or moreabortions (93.6%). Mean parity for the entire groupwas 2.4 ± 0.24 (SEM) and there were no differences inthe mean parity for each of the therapy groups whencompared to the untreated women (Table 3). Age atbirth of the first term child ranged from 19 to 29 yearsfor the 44 parous women, with a mean age at first termbirth of 23.9 ± 0.37 years. Sixteen of the parouswomen (36.4%) breast-fed one or more of their children and although the number of untreated womenwho nursed (20%) tended to be lower than in any ofthe hormone therapy groups, none of these differenceswere significant. There were 17 patients (32.1%) whogave a positive family history of breast cancer (mother,aunt, grandmother) without any significant differencesbetween the various groups. Weight ranged from 46.7to 109.8 kg, with a mean weight of 66.1 ± 1.87 kg.Although there were no differences pairwise, the meanweight of the untreated women (73.9 ± 4.34 kg) washigher with a significant main effect than that of theother four groups combined (P ≤ .04). Systolic bloodpressure ranged from 94 to 184 mmHg, with a mean of130.4 ± 2.48, and no significant differences wereobserved between any of the groups. Diastolic bloodpressure ranged from 50 to 108, with a mean of 80.2 ±1.45. The mean diastolic blood pressure of the estrogenusers (75.1 ± 2.12) was significantly lower than that ofthe untreated women (85.4 ± 2.19), with P ≤ .03, butno other differences were observed. There were six ofthe 53 women that were hypertensive (11.3%), and five of these were receiving antihypertensive medicationwhen their breast cancer was diagnosed. There wasone patient with hypertension in each of the followinghormone therapy groups: estrogen-progestogen users, estrogen users, and estrogen vaginal cream users.Three of the 18 untreated women were being treatedwith antihypertensive medication.
Table 3. Other Factors That May Predispose Toward or Protect From Breact Cancer
During the seven years of prospective study from1975 through 1981, 53 breast cancers were diagnosed inthe patients registered in the hormone use survey(Table 4). There were 5563 patients registered for atotal of 37,236 patient-years of observation, so theoverall incidence of breast cancer was 142.3:100,000women per year. The lowest incidence of mammarymalignancy was observed in the estrogen-progestogenusers, with eight patients diagnosed with breast cancerduring 11,895 patient-years of observation, for anannual incidence of 67.3:100,000 women. During15,606 patient-years of observation, there were 22breast cancers in the estrogen users, for an incidence of 141.0:100,000. With 18 breast cancers during 5258 patient-years of observation, the incidence in the untreated women was 342.3:100,000. Statistical analysisof these data are given in Table 5, also comparing thepresent incidence rates with those expected accordingto the National Cancer Institute SEER (1980) data.[3] The NCI SEER data were reported toward the end of thepresent study and represents cancer incidence in theUnited States from 1973 to 1977.
Table 4. Incidence of Breast Cancer at Wilford Hail LiSAF Medical Center: 1975 to 1981
Table 5. Statistical Analysis of Breast Cancer Incidence Data
The mean ages (± standard deviation) for each ofthe patient groups are as follows: estrogen-progestogen users = 55.2 ± 5.12 years; estrogen users = 57.1rt 6.89 years; estrogen vaginal cream users = 59.4 ±7.66 years; progestogen or androgen users = 53.4 ±5.62 years; untreated women = 58.8 ± 8.16 years; andtotal population = 56.8 ± 6.75 years. The expectedincidence of breast cancer in this age group, accordingto the Third National Cancer Survey, is 188.3:100; forages 55 to 59, according to the NCI SEER data, it is 229.2:100,000. The incidence of breast cancer in theestrogen-progestogen users (67.3:100,000) was statistically significantly lower than the untreated women(342.3:100,000) with P ≤ .01. The difference betweenthe estrogen-progestogen users (67.3:100,000) and theestrogen users (141.0:100,000) was not significant butdoes indicate a trend (P ≤ .08). The incidence of breastcarcinoma in the estrogen users (141.0:100,000) wassignificantly lower than that of the present untreatedgroup (342.3:100,000) with P ≤ .01. The incidence ofbreast cancer in the estrogen vaginal cream users(95.8:100,000) was significantly lower (P ≤ .05) whencompared with the untreated group (342.3:100,000),When stratified for age and the observed incidencewas compared to the expected incidence according tothe NCI SEER data, the only significantly lower incidence of breast cancer was observed in the estrogenprogestogen users, with a relative risk of 0.3 (95%confidence interval, 0.1 to 0.8). Although the incidencein the estrogen users was not significantly lower fromthat expected according to the NCI SEER data (relativerisk = 0.7, 0.5 to 1.1), there was a trend in thatdirection. The incidence of breast malignancy wassignificantly higher in the untreated group when compared with the NCI SEER data (relative risk = 1.4, 1.1to 1.9). The mean age (61.3 ± 2.16 years) and the meanweight (73.4 ± 4.3 kg) of the untreated women washigher than the four therapy groups combined. Thiswould account for some of their higher incidence ofbreast cancer because the expected incidence for age 60 to 64 is 251.4:100,000. However, taking into accountthese variables for age and weight, the statisticalcomparisons between the various groups remain unchanged and are still significant.
The data from the prospective study are showngraphically year by year in Figure 1. Although therewas considerable variation each year, the incidence inthe estrogen-progestogen users varied but little duringthe first four years, with a low of 80.6:100,000 in 1975to a high of 169.5:100,000 in 1976. Seven of the breastcancers were diagnosed in the first four years of thestudy, with one additional carcinoma of the breastdetected during the last three years (1980). The twomammary malignancies observed in progestogen users(P) occurred in 1978 and 1979. The incidence of breastcancer in the estrogen users remained relatively constant throughout the seven years with a low of110.3:100,000 in 1979 and a high of 173.4:100,000during 1981. Two of the three breast cancers found inthe estrogen vaginal cream users (EVC) occurred in1979 and the other in 1977. The incidence of mammary malignancy in the untreated group was also quitevariable, from a low of 227.3:100,000 in 1975 to a highof 633.9:100,000 in 1979. Overall, the incidence ofbreast carcinoma increased slightly from 143.4:100,000in 1975 to a high of 183.8:100,000 in 1978, followed bysignificant decreases to 104.2:100,000 in 1980 and110.4:100,000 during 1981.
Figure 1. Incidence of breast cancer year by year from 1975 to 1981 in the hormone users compared with the untreated women. P ~ progestogen users; EVC = estrogen vaginal cream users.
More detailed data on the eight estrogen-progestogen users diagnosed with breast cancer is presented in Table 6. In six of the eight patients, estrogens had beenused for two to 13 years before the progestogen wasadded. Only one patient (case #38) had the estrogenand progestogen started at the same time, and theprogestogen was given for only five days each monthfor five years before carcinoma was detected, tn oneother patient (case #18), Oracon, a sequential birthcontrol pill containing only five days of progestogen each cycle, was used for eight years, and then theestrogen-progestogen combination was given for 15months before breast cancer was diagnosed. Progestogens were prescribed for five days each month in onepatient, seven days in three patients, and ten days infour of these women for six months to five years beforemammary malignancy was detected. Only two of theeight women had used the progestogen for longer thantwo years, with a mean use duration of 1.8 ± 0.50years.
Table 6. Breast Cancers in Estrogen-Progestogen Users
The concept of adding progestogens to estrogenreplacement therapy was introduced at Wilford HallUSAF Medical Center in 1971. Figure 2 compares thenumber of estrogen and estrogen-progestogen treatedwomen to the incidence of breast cancer for the tenyears from 1972 through 1981. With increased estrogenuse from approximately 1320 patients in 1972 to 3940estrogen-treated women in 1975, there was no increasein the incidence of mammary malignancy. The apparent decline in the incidence of breast cancer from189.4:100,000 in 1972 to 143.4:100,000 during 1975 wasnot statistically significant. However, with ever increasing progestogen usage from approximately 9.1%of the estrogen users in 1972 to 51.1% of the estrogen users during 1981, a significant decrease in the incidence of breast cancer occurred in the ninth and tenthyears of study, with an incidence of 104.2:100,000 in1980 and 110.4:100,000 during 1981.
Figure 2. Comparison of the number of estrogen and 5 estrogen-progestogen treated women and the incidence of § breast cancer from 1972 to 1981. Dotted lines = estimated g data from computerized pharmacy records; solid lines = w data from prospective study.
Discussion
It required seven years of prospective study and tenyears of ever increasing progestogen use to indicatethe protective effect of added progestogen to estrogenreplacement therapy upon the breast, although earlierreports indicated a trend in this direction.[13,14] Apparently, it takes long-term progestogen use to reduce therisk of breast cancer in postmenopausal women. Thereduction in the risk of endometrial cancer from addedprogestogen was confirmed in the first few years of thepresent study and also has been shown in severalother studies.[12-17] Progestogens physically shed theendometrium each month, leaving behind fewer cellsand glands to continue proliferation that may eventually lead to hyperplasia and neoplasia after severalyears. The increased risk of endometrial cancer fromunopposed estrogen therapy is nullified within two tothree years after discontinuing estrogen therapy.[11]
Additional actions of both natural progesterone andsynthetic progestogens are important in addition to thephysical shedding of the endometrium. Progestogensdecrease estrogen receptors in endometrial cells andinduce estradiol dehydrogenase activity, which is themechanism where the cells metabolize estrogens.[17] Because breast cells are not cyclicly shed by progesterone, the probable protective mechanism of progestogens is most likely at the intracellular level throughchanges in receptors and enzymatic activity. The presence of both estrogen and progesterone receptors inbreast cancer tissue is related to the longer disease-freeinterval, decreased mortality, and more predictive of afavorable response to endocrine manipulation.[4]
It may be that progestogens alone, without adequateendogenous estrogens, are not as protective fromcarcinoma of the breast. Estrogens induce progesterone receptors so adequate estrogen may have to bepresent to induce progesterone receptors before progestogens can exert their protective effect. Althoughlower, there was no significant difference in the incidence of breast cancer in the estrogen-progestogenusers (67.3:100,000) as compared with the progestogenusers (148.5:100,000). However, two breast cancerswere observed in the progestogen users whereas noendometrial cancers have occurred to date. In thisrespect, carcinoma of the breast may differ from adenocarcinoma of the endometrium in that both progestogens alone and added progestogen to estrogen therapyseem to protect from endometrial cancer while it takesa combination of estrogen-progestogen therapy toreduce the risk of breast cancer.
If the ever-increasing incidence of breast cancer byage is closely examined, the role of female sex steroidsbecomes somewhat clarified.[1,2] The greatest increasein breast cancer is between the late 30s and early 50s.What is happening at this time in a woman's life isdeclining production of estrogens from the ovaries asmenopause is approached. Perhaps more important isthe fact that more women become anovulatory in thepremenopausal years, resulting in an abrupt cessationof the cyclic progesterone levels that had been presentthroughout the reproductive years. The incidence ofbreast cancer continues to increase throughout thepostmenopausal years when estrogen levels are lowerbut not absent. However, few postmenopausal women, if any, produce progesterone. If unopposed estrogens were the cause of breast cancer, the incidence ofthis malignancy would peak in the 50s and 60s anddecline thereafter, as does the incidence of endometrialcancer. Whatever the role of female sex steroids ascofactors or predisposing factors for mammary malignancy, progesterone deficiency seems to be more important than unopposed estrogen. The estrogen window hypothesis states that unopposed estrogens,caused by progesterone deficiency or luteal dysfunction, may provide a state favorable to the induction ofbreast cancer by carcinogens in the susceptible mammary gland.[20]
There is other evidence that progesterone deficiencymay increase the incidence of breast cancer. In a longterm follow-up of a group of infertility patients, thosewith progesterone deficiency had 5.4 times the risk ofpremenopausal breast cancer compared with womenin the nonhormone group (those whose infertility wascaused by other factors).[21] Women in the progesteronedeficiency group also experienced a tenfold increase indeath from all malignant neoplasms compared withthe nonhormone group. However, the incidence ofpostmenopausal breast cancer did not differ significantly between the two groups. However, anotherstudy of long-term progesterone deficiency did find anincreased risk of postmenopausal breast cancer.[22] Chronic anovulation increased the risk of endometrialcancer fivefold, and the relative risk of breast cancerafter the age of 55 years was 3.6. In the present 53patients with carcinoma of the breast, nulliparity wasperhaps a little higher than expected (17.0%), and theabortion rate also seemed high (19.8%). More important, 44 of 47 patients who had conceived had one ormore abortions (93.7%). This could be indicative oflong-term luteal dysfunction, at least in some of thepatients.
Some of the potentially predisposing and protectingfactors for breast cancer are related to progesteronedeficiency while others are not. Protecting factors suchas nulliparity and an early term pregnancy relate toperiodic high progesterone production and ovulationwith normal cyclic progesterone levels.[4,23 24] Nulliparity, infertility, and chronic anovulation, all progesterone deficiency states, not only predispose to endometrial cancer but also to breast cancer. Obesity is apredisposing factor for breast cancer and may also leadto anovulation. The risk factors for endometrial andbreast cancer are similar, and in fact, there appears tobe some association between these two cancers. Women with endometrial malignancy are more prone todevelop cancer of the breast, ovaries, and large intestine. Patients with breast cancer may have a secondcancer of another organ, including the uterus, ovaries,and colon. Certainly, there are other risk factors forbreast cancer that are probably not related to progesterone deficiency. These include a strong family history, early menarche, and late menopause. Protectingfactors not related to progesterone deficiency includebreast feeding, late menarche, early menopause, andbilateral oophorectomy before age 40, which woulddiminish estrogen levels but deplete progesterone proauction. Therefore, breast cancer must be a multifactorial disorder in which genetic traits, endocrine relationships, oncogenic factors such as viruses, andenvironmental factors such as chemical carcinogens allhave a role.
In the ten-year double-blind study of Nachtigall et al,[15] four breast cancers were detected in the 84 placebo users and none in the 84 estrogen-progestogenusers, which was statistically significant (P ≤ .05). Several studies indicate that oral contraceptives reducethe risk of benign breast disease and there has been noevidence that birth control pills increase the risk ofbreast cancer.[25-30] Some studies have not found anincreased risk of breast cancer in oral contraceptiveusers, while others observed lower rates of malignancyin those taking birth control pills. Long-term studies oflarge numbers of women have failed to incriminateestrogen replacement therapy as a risk factor for breastmalignancy. Of the six groups that have studied bothendometrial and breast cancer among estrogen users, amodest association was noted between estrogen therapy and endometrial cancer; however, in every instancethe association between estrogens and breast cancerwas considerably less.[5-7,13,15,16]
The present study found a decreased incidence ofbreast cancer in the patients that used estrogens only.Although the risk was significantly lower than in theuntreated group, there was no difference when compared with the NCI SEER data; however, because therelative risk was 0.7, this indicated a trend in thatdirection. Hoover et al[5] observed an insignificantlyincreased risk of breast cancer (relative risk = 1.3) frompostmenopausal estrogen use. In a later study from thesame patient population, a decreased risk of breastcancer was found in estrogen users by Bland et al.[31] hev followed 405 postmenopausal women (mean age 54,7 years) for three to more than 28 years with serialmammography, using xeromammography in the lateryears of the study. The 206 estrogen-treated womenhad received therapy for a minimum of 18 months(mean 6,5 years) and had a lower incidence of breastcancer than the untreated women. They concludedthat long-term estrogen replacement does not significantly alter mammographic parenchymal patterns andthat estrogen use does not increase the risk of breastcancer. Hammond et al[16] had four cases of breastcancer among 301 estrogen-treated women followed for five or more years, and four cases in the 309 untreated women.
Only two studies have observed any significantlyincreased risk of breast cancer from estrogen therapy. In neither was the risk increased in the total studyPopulation, but rather in subgroups of estrogen users,In the first by Ross et al,[6] a slightly increased risk was reported in a small subgroup of estrogen users withintact ovaries who had received a total dose of greaterthan 1500 mg (relative risk = 2.5; P ≤ .02). Theyobserved a lower relative risk of breast cancer in fourother subgroups of estrogen users: those with previousoophorectomy, and those who had ever undergoneestrogen treatment, relative risk — 0.8; prior oophorectomy with total milligram dose > 1500, relative risk =0.9; prior oophorectomy with total milligram dose > 1500, relative risk = 0.7; and intact ovaries with totalmilligram dose < 1500, relative risk = 0.9. In the otherstudy by Jick et al[,7] no association was observedbetween current estrogen use and carcinoma of thebreast in women with previous hysterectomy (relativerisk = 1.1). Only slightly over 50% of the women alsohad oophorectomy, and the rate of breast cancer inthose with and without oophorectomy was similar.Although oophorectomy may lower the risk for breastcancer, it is difficult to understand why hysterectomywith conservation of the ovaries would also lower therisk. In women with a natural menopause and intactuterus, a positive association was found between current estrogen use and breast cancer (relative risk =3.4). This association varied by age, with a relative riskof 10.2 in women aged 45 to 54 years, compared with arelative risk of 1.9 in those aged 55 to 64 years. Thework of both the Ross and Jick groups can be criticizedbecause they limited the study population to age 75and age 65, respectively, yet the incidence of breastcancer continues to increase with each five-year increment in age. Because more women are treated withestrogens in the early menopausal years than after age60, this alleged increased risk of breast cancer probablyreflects greater estrogen use in this age group and maynot be a true association at all.
In conclusion, estrogen replacement therapy doesnot increase the risk of breast cancer and may possiblyafford some protection. Added progestogens to estrogen therapy significantly reduce the risk of mammarymalignancy, so should also be given for ten days eachmonth to patients who have had a hysterectomy.Unlike endometrial cancer, it may be that progestogens only do not help to prevent cancer, so consideration should be given for combinations of estrogenprogestogen replacement therapy when indicated forpostmenopausal hormone therapy.
References
Cutler SJ, Young JL. Jr: Third National Cancer Survey: incidencedata. Natl Cancer Inst Monogr 41, 1975, p 110
National Cancer Institute: Surveillance, epidemiology, and endresult (SEER). Bethesda, Biometry Branch of the National CancerInstitute, 1980, p 47
American Cancer Society: Cancer Fact* and Figures, 1983. NewYork, American Cancer Society, 1982, p 7
Alpert U: Risk and prognostic factors in breast cancer. TheFemale Patient 7:55, 1982
Hoover R, Gray I.A Sr, Cole P, et al: Menopausal estrogens andbreast cancer. N Engl J Med 295:401, 1976
Ross RK, Paganini-Hill A, Cerkins YR, et al: A case control studyof menopausal estrogen therapy and breast cancer. JAMA243:1639, 1980
Jick H, Walker AM, Watkins RN, et al: Replacement estrogensand breast cancer. Am ) Epidemiol 112:586, 1980
Ziel HK, Finkle WD: Increased risk of endometrial carcinomaamong users of conjugated estrogens. N Engl J Med 293:1167,1975
Smith DC, Prentice R, Thompson DJ, et al: Association ofexogenous estrogen and endometrial carcinoma. N Engl J Med293:1164, 1975
Gray LA Sr, Christopherson WM, Hoover RN: Estrogens andendometrial carcinoma. Obstet Gynecol 49:385, 1977
Hulka BS, Fowler WC Jr, Kaufman DG, et al: Estrogens andendometrial cancer: Cases and two control groups from NorthCarolina, Am J Obstet Gynecol 137:92, 1980
Gambrell RD Jr, Massey FM, Castaneda TA, et al: Use of theprogestogen challenge test to reduce the risk of endometrialcancer. Obstet Gynecol 55:732, 1980
Gambrell RD Jr: Role of hormones in the etiology and preventionof endometrial and breast cancer. Acta Obstet Gynecol ScandSuppl 106:37, 1982
Gambrell RD Jr: Clinical use of progestins in the menopausalpatient. J Reprod Med 27:531, 1982
Nachtigall LE, Nachtigall RH, Nachtigall RB, et ah Estrogenreplacement 11: A prospective study in the relationship to carcinoma and cardiovascular and metabolic problems. Obstet Gynecol54:74, 1979
Hammond CB, Jelovsek FR, Lee KL, et al: Effects of long termestrogen replacement therapy. II. Neoplasia. Am j Obstet Gynecol 133:537, 1979
Whitehead MI, Townsend PT, Pryse-Davies J, et al: Effects ofestrogens and progestins on the biochemistry and morphology ofthe postmenopausal endometrium. N Engl J Med 305:1599, 1981
Mantel N: Chi-square tests with one degree of freedom. Extensions of the Mantel-Haenszel procedure, Am Stat Assoc J 58:690,1963
Miettinen OS: Estimability and estimation in case-referrant studies. Am J Epidemiol 103:226, 1976
Korenman SG: The endocrinology of breast cancer. Cancer 46-:874, 1980
Cowan LD, Gordis I., Tonascia JA, et al: Breast cancer incidencein women with a history of progesterone deficiency. Am JEpidemiol 114:209, 1981
Coulam CB, AnnegC'rs JF: Chronic anovulation may increasepostmenopausal breast cancer risk. JAMA 249:445, 1983
Kirschner MA: The role of hormones in the etiology of humanbreast cancer. Cancer 39:2716, 1977
Vorheer H, Messer RH: Breast cancer: Potentially predisposingand protecting factors. Am J Obstet Gynecol 130:335, 1978
Keifer WS, Scott JC: A clinical appraisal of patients followinglong-term contraception. Am j Obstet Gynecol 122:446, 1975
Vessey MP, Doll R, Jones K: Oral contraceptives and breastcancer. Lancet i;941, 1975
Ory H, Cole P, MacMahon 13, et al: Oral contraceptives andreduced risk of benign breast disease. N Engl.) Med 294:419, 1976
Drill VA: An overview' of studies on estrogens, oral contraceptives and breast cancer. Prog Drug Res 25:159, 1981
Kay C: Breast cancer and oral contraceptives: Findings in RoyalCollege of General Practitioners' study. Br Med j 282:2089, 1981
Ramcharan 5, Pellegrin FA, Ray R, et al: The Walnut CreekContraceptive Drug Study: A prospective study of the side effectsof oral contraceptives, Vol III. Bethesda, DHWS (NIH), 1981
Bland KI, Buchanan JB, Weisberg BF, et al: The effects ofexogenous estrogen replacement therapy of the breast: Breastcancer risk and mammographie parenchymal patterns. Cancer45:3027, 1980