Autoimmune Disease is Increased in Women With Primary Ovarian Insufficiency.

After a thorough search of diagnostic codes and chart review at the University of Utah and Intermountain Healthcare, 610 women were identified with a diagnosis of POI. The average age (± SD) at the time of POI diagnosis was 32.7 ± 7.4 years, and the average age at the time of the study was 48.3 ± 11.9 years (range 14-93 years). Race/ethnicity of the women were self-reported. The distribution was predominantly White (78.7%), White Hispanic (10.5%), Asian (1.6%), and multiple races (3.8%). The remaining encompassed Native American, Pacific Islander, Black, and unknown. Of the 610 women, 410 had 3 generations of genealogical data available in the UPDB. These 410 women had 2271 first-degree relatives, 5683 second-degree relatives, and 11 697 third-degree relatives. Of the 610 women, we identified 153 subjects with any autoimmune disease (25%). Women with POI had an increased risk for several autoimmune diseases compared to population rates of autoimmune diseases ( Table 1 ). Of the autoimmune diseases identified, most of the women had hypothyroidism (79% of the women with autoimmune disease). Fewer than 10% of those with autoimmune disease had each of the following: RA, vitiligo, celiac disease, T1DM, psoriasis, adrenal insufficiency, and SLE. Of the total with autoimmune disease, 117 women (78%) had 1 autoimmune disease and 36 women (22%) had multiple autoimmune diagnoses. Among the subjects with multiple autoimmune diseases, there were ≤10 women with adrenal insufficiency. All women with adrenal insufficiency met the criteria for autoimmune polyglandular syndrome type 1 if there was an associated diagnosis of hypoparathyroidism or autoimmune polyglandular syndrome type 2 with associated disorders including hypothyroidism, vitiligo, T1DM, and celiac disease. The remaining subjects had various combinations of autoimmune diseases, including APS types 3 and 4, and other combinations that did not fit into APS classifications ( Table 2 ). The age at diagnosis for most of the autoimmune diseases was as expected based on population ages and the literature (Supplementary Table S1) ( 17 ). Autoimmune disease identified in women with POI (n = 610) The observed number of cases with autoimmune disease, expected number based on population rates, and the relative risk. Based on the Utah Population Database regulations for oversight by the Utah Resource for Genetic and Epidemiologic Research, observed counts < 10 were not listed to protect the identity of the patients. No women with POI were found to have Crohn’s disease, systemic sclerosis, or Behcet disease. Abbreviations: CI, confidence interval; POI, primary ovarian insufficiency. Bonferroni multiple testing correction. * P -value < .0028. Observed Number of Women with POI and more Than 1 Autoimmune Diseases (n = 36) Abbreviations: APS, autoimmune polyglandular syndrome; MS, multiple sclerosis; POI, primary ovarian insufficiency; RA, rheumatoid arthritis; SLE, systemic lupus erythematosus We looked for confirmatory antibody testing for autoimmune disease by chart review. Of the total number of women with POI, only 42 had 21 hydroxylase antibody testing or adrenal insufficiency. In those women with POI and antibody testing who did not have a known diagnosis of adrenal insufficiency before or at the same time as the diagnosis of POI, only 2 women had positive 21 hydroxylase antibodies, but they were only minimally positive. One subject was negative on repeat testing, and the other had a different known genetic diagnosis. For other autoimmune diagnoses, 23 women with hypothyroidism had antibody testing and 9 (39%) were found to be positive; 8 women with RA had antibody testing and 4 (39%) were positive; 2 women, each, with T1DM and celiac disease had antibodies tested and 1 (50%) was positive for each; 4 women with SLE had antibody testing and 2 (50%) were positive; and for myasthenia gravis, 1 woman was tested with 100% positive. We separately investigated the risk of the autoimmune conditions in the 410 women with POI and 3 or more generations of genealogical data. We observed similar increased risks of the same autoimmune diseases with POI as in the complete POI cohort. Two exceptions were SLE, which was not significant after correcting for multiple testing, and myasthenia gravis, which was found to be associated with POI ( Table 3 ). There was no increased risk of autoimmune disease in the relatives of women with POI after correction for multiple testing ( Table 4 ). Autoimmune disease identified in the subset of women with POI with 3 or more generations of genealogical data (n = 410) The observed number of cases with autoimmune disease, expected number based on population rates, and the relative risk. Based on regulations of the Utah Population Database oversight by the Utah Resource for Genetic and Epidemiologic Research, observed numbers were listed as < 10 to protect the identity of the patients. No women were found to have Crohn's disease, systemic sclerosis, Behcet's disease, or primary biliary cirrhosis. Abbreviations: CI, confidence interval; POI, primary ovarian insufficiency. Bonferroni multiple testing correction. * P -value < .0028. Relative risk of autoimmune disease among first-degree, second-degree, and third-degree relatives of women with POI There was no significant increase in the relative risk of any autoimmune disease. For relative groups missing any of the autoimmune diseases, there were no cases identified. Abbreviations: CI, confidence interval; POI, primary ovarian insufficiency. Bonferroni multiple testing correction. * P -value < .0028.

Our data demonstrate autoimmune disease in 25% of women with POI and an increased relative risk for multiple autoimmune disorders in women with POI, confirming previous findings ( 18 ). Women with multiple autoimmune disorders presented with components of autoimmune polyglandular syndrome types 1 through 4. Our data do not demonstrate an increased relative risk for the same autoimmune disorders in family members of women with POI. These data suggest that other factors in addition to shared POI and autoimmune genetic risk play a role in the increased autoimmune disease burden in POI. Our data demonstrate an increase in adrenal insufficiency in the context of APS types 1 and 2, as expected based on the known relationship between adrenal and ovarian autoimmunity ( 19 ). APS type 1 is characterized by a triad of Addison's disease, chronic mucocutaneous candidiasis, and hypoparathyroidism, due to mutations in the autoimmune regulatory gene, AIRE ( 16 ). APS type 2 is a cluster of 2 or more endocrine diseases, commonly thyroid disease and T1DM, with a more complex inheritance pattern ( 16 ). The prevalence of clinical adrenal insufficiency, ≤ 1.6% in our dataset of women with POI, is similar to that reported in previous studies demonstrating 0.2% of women with POI and adrenal insufficiency ( 6 , 18 ). In addition, approximately 4.5% to 5% of women with POI will have adrenal autoimmunity without current adrenal insufficiency ( 3 , 20 ). Many of our subjects did not have available testing for adrenal cortical or 21-hydroxylase antibodies. Therefore, our number with adrenal autoimmunity is likely an underestimate at 2.9%. In patients with autoimmune adrenal failure, antibodies to the adrenal cortex and to steroid cells have been identified by indirect immunofluorescence on fixed tissue samples and best discriminate autoimmune POI from POI of other etiologies ( 7 , 19 ). Autoimmune adrenal insufficiency or the presence of adrenal autoantibodies are the only autoimmune diseases or antibody manifestations that have been documented in association with histologically proven ovarian autoimmune oophoritis that initially targets the steroid-producing theca cells ( 7 , 8 ). These infiltrates eventually spill over, destroying the surrounding tissue and resulting in ovarian insufficiency ( 8 ). As demonstrated previously, the most common autoimmune disease was hypothyroidism, with a prevalence of 19.8%, accounting for a component of autoimmune disease in 79% of the women with POI and autoimmune disease in our cohort. Hypothyroidism has been found in 12% to 15% of patients with POI in previous studies ( 7 , 18 , 21 ), slightly lower than in the current data. Thyroid autoimmunity has been referred to as the presence of antibodies to thyroid peroxidase, TSH receptor, and thyroglobulin ( 22 ). Previous studies did not report an association between thyroid autoimmunity and fecundity or ovarian reserve ( 23 , 24 ), suggesting that the relationship is not causal for POI despite the association. The prevalence of T1DM was 2.0% among women with POI in our study, similar to previous findings ( 7 ). The majority of the subjects with T1DM also had hypothyroidism and POI, which has been categorized as APS type 3 ( 16 ). Women with T1DM have been found to have lower anti-Mullerian hormone levels, suggesting decreased ovarian reserve ( 25 ). The mechanism of ovarian insufficiency could be related to microvascular disease or glycosylation rather than an autoimmune mechanism itself ( 26 ). Concomitant hypothyroidism does not appear to be an important causal factor for POI, as discussed earlier. In addition to endocrine autoimmune disease, we found an increased risk for nonendocrine autoimmune diseases including RA, SLE, psoriasis, vitiligo, myasthenia gravis, and celiac disease. In some cases, these nonendocrine autoimmune diseases were found as components of APS types 2 or 3 or APS type 4 in combination with thyroid disease ( 16 ). There has been an established relationship between POI and both RA and SLE. Our study demonstrated a prevalence of 2.3% for RA and ≤1.6% SLE in women with POI. Of note, cyclophosphamide therapy for severe SLE can cause POI with high and cumulative doses ( 27 , 28 ). However, we removed patients with cyclophosphamide treatment from our cohort after our extensive chart review. Anti enolase and anti corpus luteum antibodies have been identified as ovarian targets in RA and SLE; however, these antibodies have not been reproducibly associated with POI over time and there is no clear relationship between these autoantibodies and an autoimmune infiltrate in the ovary ( 7 , 29 , 30 ). Ovarian vasculitis has also been described, with pathologic evidence described in SLE, and could play a role in ovarian damage ( 31 ). Finally, Mendelian randomization suggests a causal genetic relationship between SLE and POI, but replication is needed as the power of the study was likely inadequate ( 32 ). Overall, the relationship between RA and SLE and POI remains undetermined. Similar to previous findings, psoriasis is also increased in women with POI in the current study ( 2 ). Psoriasis has been documented as a component of APS types 2 to 4 ( 16 ) but has not been reported as frequently in association with POI. Compared to other autoimmune diseases, the underlying mechanism causing psoriasis was previously thought to be distinct, without evidence for autoantibodies or autoreactive T-cells ( 33 ). However, more recent data suggest that T cell self-reactivity is a central component of psoriasis pathogenesis, although there are still no associated autoantibodies ( 33 ). The differences in the spectrum of immune mechanisms and clinical presentations in psoriasis compared to other autoimmune disease mechanisms could provide additional insights into the relationship with POI. We did not find familial autoimmune disease in families of women with POI after correction for multiple testing. Autoimmune disease is highly heritable, as demonstrated by higher concordance between monozygotic (13-40%) compared to dizygotic twins for many autoimmune diseases ( 11-13 ). Genome-wide association studies have demonstrated complex genetic risk for autoimmune diseases, suggesting that there should be some aggregation in large families like those we studied ( 14 ). Previous family studies have demonstrated co-occurrence of autoimmune disease in family members, with the strongest evidence for T1DM, Crohn's disease, psoriasis, and multiple sclerosis ( 34 ). However, the familial cases in other diseases may be few, as there are less than 2% of women with SLE who have relatives with SLE, for example ( 13 , 15 ). Our results may be related to small numbers, which is a problem with rare autoimmune diseases, particularly because we studied them as a secondary entity in our POI families and were not studying autoimmune disease as our entry into the analysis. We may have missed cases if a diagnosis was not made or was documented in a different health system. Developing autoimmune disease also depends on environmental exposures ( 35 ). Differences in environmental exposures in women with POI compared to relatives could be a cause, although the environment may be shared especially for first-degree relatives. It is also possible that differences in sex steroid levels, particularly in estrogens and androgens, play a role in the differing disease prevalence in women with POI compared to their family members ( 15 ). In particular, changes in estradiol levels can exacerbate SLE, while androgens suppress autoimmunity in experimental models ( 15 , 17 ). It is also possible that familial aggregation may be more complicated, with 1 autoimmune disease found in a woman with POI and different autoimmune diseases in family members. Genetic studies demonstrate overlap in genetic loci among several autoimmune diseases, and at least 40% of genetic risk loci influence more than 1 autoimmune disease ( 14 ). For example, loci conferring risk for more than 1 autoimmune disease include the STAT4 locus for RA and SLE; the CD289-CTLA4 locus for T1DM and celiac disease or RA; the ICAM3 locus for RA with SLE; the TYK2 locus for RA with SLE, T1DM, and inflammatory bowel disease; and the RGS1 locus for T1DM and multiple sclerosis and celiac disease ( 14 ). In line with gene loci overlapping for different autoimmune diseases, a familial relationship among 2 or more different autoimmune diseases has been demonstrated in some studies ( 36 ). We did not examine these more complicated autoimmune disease relationships in our families. Strengths of this study include the careful case definition of POI at a population level validated with chart review and the extensive genealogy database. Our results are generalizable to other northern European populations. Although 10.4% of these women had Hispanic ethnicity ( 37 ), our results may not be generalizable to other races and ethnicities. Other limitations include our inability to identify all autoimmune conditions in the relatives, although we did use data from the 2 largest health care systems in Utah. We may also miss autoimmune disease based on miscoding and miss antibody testing based on the availability of data in the medical records. Our data demonstrate an increased relative risk for several autoimmune diseases in women with POI compared to the general population. The absence of increased risk in family members does not mean that there is no genetic predisposition to autoimmune disease in POI, but it suggests epigenetic, environmental, and/or hormonal risk may increase the likelihood of autoimmune development in women with POI. These findings support clinical practice to expand screening for autoimmune diseases in women with POI as already recommended for hormones and antibodies associated with adrenal and thyroid disease.

We identified women with POI using electronic medical records at the University of Utah Health and Intermountain Healthcare from 1995 to 2022. The cases of POI were obtained through International Classification of Disease (ICD) codes: ICD-9 (256.3, 256.31, 256.39) and ICD-10 codes (E28.3, E28.31, E28.39, E28.310, and E28.319) and electronic medical record notes indicating POI diagnoses and/or lab values (elevated FSH > 20 IU/L or anti-Müllerian hormone <0.08 ng/mL in a woman under the age of 40 years at the time of the laboratory draw). Subjects with the following current procedural terminology codes were excluded: hysterectomy, oophorectomy, endometriosis with pelvic surgery, pelvic radiation or chemotherapy before the diagnosis of POI, and Turner syndrome (ICD-9 758.6 and ICD-10 Q96). Those identified with POI ICD codes were verified through individual chart review by a medical or reproductive endocrinologist (C.K.W. or L.E.V.) for inclusion. Validation factors included lab values (FSH and anti-Müllerian hormone levels), diagnosis details, POI signs, and the type of physician making the diagnosis ( 10 ). Medical record numbers for women with POI were converted to Utah Population Database (UPDB) identification numbers by an independent oversight group under the requirements for use. Subsequently, UPDB IDs were linked to genealogy within the UPDB. For the familial clustering analysis, POI probands were required to have at least 3 generations of genealogy information available (proband, both parents, and 3 out of 4 grandparents). Three or more generations increases the likelihood that subjects and their families have lived in Utah for many generations and would have complete family data and health care data. The institutional review boards at the University of Utah and Intermountain Healthcare approved this study. We identified autoimmune diseases using ICD codes in women with POI and their relatives, choosing those found in previous studies and known polyglandular syndromes ( 7 , 16 ). These included adrenal deficiencies (ICD-9 255.X, ICD-10 27.X), hypoparathyroidism (252.1, E20.X), type 1 diabetes mellitus (250.X, E10.X), vitiligo (709.01, L80), celiac disease (579.0, K90.0), vitamin B12 deficiencies (281.0, D51.X), multiple sclerosis (340, G35), systemic lupus erythematosus (710.0, M32.X), Crohn's disease (555.X, K50.X), psoriasis (696.X, L40.X), ulcerative colitis (556.X, K51.X), inflammatory bowel disease (569.89), systemic sclerosis (701.0, M34.X), scleroderma (701.0, M34.X), RA (714.0, M05.X), hypothyroidism (244.8, 244.9, E03.X, E06.X), immune thrombocytopenia (287.31, D69.3), Behcet (136.1, M35.2), myasthenia gravis (358.01, G70.X), and primary biliary cirrhosis (571.6, K74.3). Autoimmune disease in women with POI was verified through individual chart review and discussed with clinical rheumatologists (J.W. and J.Z.). The relative risk of individual autoimmune diseases in women with POI and their relatives was analyzed. First-degree relatives included parents, siblings, and children. Second-degree relatives included grandparents, aunts/uncles, nieces/nephews, half-siblings, and grandchildren. Third-degree relatives included great-grandparents, great-grandchildren, and first cousins. The relative risk of a disease estimates the likelihood an individual may develop a disease if the woman has POI or is a relative of a woman with POI. The relative risk was calculated as a ratio of the observed number of a specific type of autoimmune disease for a woman with POI or a specific relative type compared to the expected number of autoimmune disease cases for a woman or a specific relative type. To obtain the expected number of autoimmune disease cases, population rates were calculated for each 5-year birth cohort represented by the POI subjects or relatives and birthplace (Utah vs outside of Utah) within the University of Utah Health and Intermountain Healthcare. The rate was defined to be the total number of cases of a specific autoimmune disease divided by the total cohort size. The number of expected cases was then calculated by summing each cohort-specific autoimmune disease risk for each individual in a set of relatives of a specific type (eg, first-degree relatives). Approximate 95% confidence intervals and exact hypothesis tests of the null hypothesis (relative risk = 1.0) were constructed assuming that the number of autoimmune disease cases found among the relatives follows a Poisson distribution. The studies performed with the UPDB are population based, which reduces the risk for sampling bias that can occur due to proband identification and oversampling from pedigrees with multiple affected members. We corrected for multiple testing for 18 autoimmune disorders ( P < .05/18 = 0.0028).