We provide evidence that serum concentrations of hCG are elevated in only 2.2% of potentially fertile female dialysis patients of reproductive age, in contrast to 39% of infertile patients. This finding has implications for the exclusion of pregnancy in surgery settings and anesthesia in general, for time-sensitive indications such as deceased donor kidney transplantation, and for the use of MPA as a part of primary immunosuppression.
Fertility is diminished in dialysis patients with a pregnancy incidence between < 1–7% [5, 25,26,27,28,29]. Menopause occurs in these women 4.5 years earlier than in healthy women and primary ovarian failure, defined as a sign of menopause before the age of 40 years, is frequent with a proportion of 14% in comparison to 0.01% in the general population [30,31,32]. Profound endocrine abnormalities leading to menstrual and fertility disorders in dialysis patients are likely the result of a defect in hypothalamic regulation of gonadotropin secretion  and in the absence of other clinical correlates, serum concentrations of hCG may be elevated in these patients. When they do conceive, advanced renal failure predisposes them to abortion, intrauterine growth restriction and preterm delivery. Intensified dialysis may improve fertility and pregnancy outcomes among dialysis patients,  but the rate of successful pregnancies is about four times greater after kidney transplantation (33/1000 female transplant recipients) but still less frequent as compared to the general population (> 100/1000 females) [35, 36]. Notably, two studies showed an increase of the pregnancy rates among dialysis patients in recent years [37,38,39].
Tacrolimus and MPA are recommended as first line therapy for prevention of kidney allograft rejection . However, MPA use in pregnancy is associated with an increased risk of miscarriage with a rate of 45–49% during the first trimester, and congenital defects, such as external ear malformation, cleft lip and palate, and abnormality of distal limbs, heart, esophagus or kidneys, which occur in 23–27% of cases. In comparison, the miscarriage rate in female solid organ recipients receiving other immunosuppression lies between 12 and 33% and the rate of congenital defects between 4 and 5%, which is comparable to 3% in the general US population . Therefore, the FDA and EMA recommend pregnancy exclusion immediately before starting with MPA therapy, eight days later, and the use of contraceptives during ongoing therapy. Furthermore, the Report by the American Society of Anesthesiologists Task Force on Preanesthesia Evaluation suggested that pregnancy testing may be offered to female patients of childbearing age and for whom the result would alter the patient’s management, because patients may present for anesthesia with early undetected pregnancy . Surgery and anesthesia on a pregnant woman may have significant implications for the fetus and the mother, and it is commonly recommended that all surgery, unless truly emergent, be postponed until after delivery to minimize the risk to the fetus . However, withholding indicated surgery from a pregnant woman as a result of fears of teratogenesis, pregnancy loss, or preterm birth would appear to be unfounded and may significantly contribute to both maternal and neonatal morbidity . Taken together, an undetected pregnancy in a dialysis patient who undergoes kidney transplantation and uses MPA as part of the immunosuppressive therapy poses an unacceptable risk to the fetus .
Pregnancy can be diagnosed or excluded by measurement of hCG in urine or serum. It is a heterodimeric glycoprotein hormone composed of an alpha- and a specific beta-subunit. The hCG alpha-subunit is identical to the alpha-subunit of LH, FSH, and thyroid-stimulating hormone (THS) and before release into circulation, the alpha- and beta-subunits are non-covalently bound. Human chorionic gonadotropin is normally secreted by the syncytiotrophoblast of the placenta, but also by trophoblastic and gastrointestinal tumors. The main functions of hCG include the maintenance of progesterone secretion from the corpus luteum until the placenta takes over this function after 6 weeks of gestation; it also stimulates gonadal testosterone secretion of the male fetus. Very small amounts in men and women primarily derive from the anterior pituitary gland [42, 43]. Some 8% of menopausal women present with elevated serum concentrations of hCG of pituitary origin > 5 mlU/ml and a higher cut-off of 14 mlU/ml is recommended for women > 55 years .
Today, serum hCG is measured by highly specific 2-site immunometric assays using antibodies specific for the beta-subunit . The urine hCG pregnancy tests detect the free beta-subunit and are less sensitive as compared to serum assays. In dialysis patients, urine tests are not recommended and not possible because of anuria, and serum hCG based pregnancy testing is reported to be unreliable due to a high rate of “false” positive results . However, due to the lack of reliable data, this suggestion is only supported by a few case reports in the literature (summarized in Table S4).
Therefore, hCG serum concentrations were examined with a highly sensitive and specific test in female dialysis patients of reproductive age. Potential fertility and infertility were diagnosed with a detailed medical and gynecological history and measurement of serum concentrations of FSH, AMH and LH. Except for two pregnant patients, 46 out of 69 non-pregnant cases (67%) were classified to be potentially fertile and 23 as infertile. Only one of the non-pregnant, potentially fertile women (2.2%) presented with an hCG serum concentration > 5 mlU/ml (case 6 in Table 5). Nine further patients with elevated hCG > 5 mlU/ml (6 to 25 mlU/ml) were considered infertile, which represented 39% of the infertile patient group. Thus, among female dialysis patients of reproductive age, 14.5% presented with elevated serum hCG, and the vast majority of them were classified as infertile. In contrast to age, potential fertility and pregnancy were independent predictors of hCG serum concentrations.
In regard to diagnostic accuracy, this study showed that the hCG cut-off of > 5 mlU/ml had a specificity of 86% for the diagnosis of pregnancy among the group of female dialysis patients of childbearing age as a whole. Using a higher cut-off of 14 mlU/ml for the subgroup of infertile patients, specificity increased to 93%. For the group of potentially fertile patients alone, specificity improved to 98%. The positive predictive value increased accordingly in all three analyses, whereas sensitivity and the negative predictive value remained at 100%. Thus, for potentially fertile dialysis patients of childbearing age a standard hCG serum concentration cut-off of ≤5 mlU/ml can be used to safely exclude early pregnancy. In case of an unknown fertility status, the ideal cut-off for the diagnosis of pregnancy in our patient population was 25 mlU/ml, which corresponds well to the cut-off suggested by Braunstein et al. . When a patient presents with an elevated hCG serum concentration, causes other than pregnancy or menopause should be considered. There may be malignancies or test interference with heterophilic antibodies, which are frequently encountered in patients with autoimmune diseases [44, 45].
Potential limitations to the study should also be considered, such as the small sample size which diminishes the generalizability of our data. Due to this, the wide confidence intervals reflect random sampling variability. However, given the importance of the outcome in light of the very low frequency of this condition, it appears to be a disproportionate effort to recruit more participants only to gradually increase precision before reporting this observation. Likewise, a confidence interval around the 100% predictive value is a result of no observations rather than a high precision. We cannot exclude the possibility of false negatives with a larger sample size. Moreover, by avoiding a case control study design, it was possible to exclude an important source of selection bias. Blinding of the index test was assured by the analysis of hCG in a department of laboratory medicine. There was no blinding of the reference standard, but we do not deem this a major source of bias for advanced pregnancy. However, this may not be the case for miscarriages. A stratum bias does not seem likely because patients were recruited from several institutions without traceable selection.