Failure of Ovarian Suppression With Gonadotropin-Releasing Hormone Analogs to Preserve Fertility: An Assessment Based on the Quality of Evidence.

Abstract

The proximate cause of menopause, whether natural or induced, is ovarian primordial follicle oocyte reserve depletion. This reserve is established inutero and cannot be replenished. Specific chemotherapeutics, particularly alkylatingagents and topoisomerase inhibitors, induce DNA double strand breaks in primordial follicle oocytes, thereby triggering apoptotic death.1 It is by this reduction of the ovarian folliclereservethatthereproductive life span is shortened andmenopause is facilitated. Ovarian failure is the last manifestation of chemotherapy-induced infertility. Hence, ovarian follicle depletion and infertility due to cancer treatments are not independent processes. In the meta-analysis by Munhoz et al,2 the authors conclude that gonadotropin-releasing hormone analog (GnRHa) suppression preserves menstruation but not fertility in premenopausalwomenwithearly-stagebreast cancer.Howdowe explain this contradiction?As the authors indicate,menstruation is not a surrogate of fertility. In fact, the commonly accepteddefinitionof premature ovarian failure (POF) is irregular periods or amenorrhea with 2 serum follicle-stimulating hormone (FSH) values greater than 40mIU/mL (to convert to IU/L,multiplyby1.0).Prior studies incorrectlyapplied thedefinition of natural menopause (ie, a retrospective absence of menses for 12months inwomen>40years) to chemotherapyinduced POF.2 Irregular menstruation is more common than amenorrhea inwomenexperiencingearlymenopause. In a recent 18-month longitudinal follow-up of women with breast cancer receiving chemotherapy, we found that approximately one-third of women with severely diminished and/or undetectable ovarian reserves by serum antimullerian hormone (AMH) levels were still menstruating.3 This brings up a major weakness of all studies that show some benefit in preserving menstruation by GnRHa. Each used different criteria and none reported on the frequency of periods and whether theprechemotherapymenstrualpatternwaspreserved.Moreover, many included women close to menopausal age. Given that at least one-third ofwomen in early POFmay bemisclassified as having normal ovarian function, the use of resumptionof periods afterGnRHa treatments as a gaugeof effectiveness of ovarian suppression is questionable. Furthermore, tamoxifen isanovarianstimulantandcanaltermenstrual functionandcause amenorrhea.Only 3 studies, 2withnegative results, excludedwomen receiving tamoxifen. Studiesmeasuring menstrual regularity postchemotherapy did not find a benefit of GnRHa in ovarian protection.4 In addition,no trial usedblindingand/orplacebo.Theparticipantswho are aware of theGnRHa treatmentmaybemore likely to report anybleeding asmenstruation. Ideally, short of using long-term fertility rates, theovariannormalcy shouldbe assessed by quantitative biomarkers. Some studies randomly measured serumFSHandestradiol, and this is not reliableunless performed on cycle days 2 to 3 as the FSH and estradiol serum values fluctuate throughout the cycle. Antral follicle count (AFC) is amore reliablemeasure,which is likewisecycledaydependent. Finally, serumAMH is thought to be themost reliableand lesscycle-dayspecificmarkerbecauseofbeingproduced from early-stage follicles. Only 2 randomized studies employed ovarian reserve markers appropriately. Demeestere et al5 used the widely accepted criterion of FSH level greater than 40 mIU/mL to determine POF in GnRHa-treated women with lymphomas after 12months. Elgindy et al,4 in addition to regularmenstrual status, followed women with breast cancer measuring AFCs andAMHlevels forup to 12months.Neither study foundabenefit from GnRHa treatments. The currentmeta-analysiswas limited to premenopausal women with early-stage breast cancer only.2 Given that the mechanismof ovarian damage by gonadotoxic agents is similar regardless of theunderlyingmalignancy type, a diagnosisbased segregation is not needed to assess the efficacy of GnRHa. The prospective studies in women with hematological malignancies, which were left out from the current metaanalysis, found ovarian suppression to be ineffective in preserving ovarian function.4 Ameta-analysis doesnot remove flaws, andas the authors indicated, amajority of patients originated from3studieswith the aforementioned inherent weaknesses. In the study by Badawy et al,6 receptor status, tamoxifen use, and the definition for resumption of menses were not mentioned. Whether this studywasactually randomizedwasalsoquestioned.7 Inanotherstudy,8authorsobservedhigherpregnancyrateswithGnRHa, thoughthiswasnot theprimaryoutcomemeasure.When recalculated based on those who were attempting pregnancy and not the total number of women in each group, there was nostatisticaldifference.Moreover, attesting to thepositivebias amongwomen receivingGnRHa, there appeared to be a trend, though not statistically significant, for a higher number of women attempting pregnancy in the GnRHa group (P = .13). In a recent meta-analysis5 that included 10 eligible trials and907womenwith theprimaryoutcomedefinedas theproportion of women with resumed ovarian function at the longest follow-upafter the endof chemotherapy,GnRHa cotreatment did not favor ovarian function resumption (320/468 Related article page 65