1. IntroductionWith the expanding use of immune checkpoint inhibitors (ICIs) across cancer types and their potential for long-term disease control, their use in pregnancy may become increasingly common [1]. Immune checkpoint pathways involving the programmed cell death-1 protein (PD-1), its ligands (programmed cell death-ligand 1 and 2, PD-L1 and PD-L2), and the cytotoxic T-lymphocyte-associated protein-4 (CTLA-4) play a physiological role in maintaining maternal immune tolerance to the developing fetus [2,3]. It is noteworthy that the PD-L1 expression on T cells at the maternal–fetal interface increases during the progression of pregnancy to prevent fetal in-utero rejection [4,5,6]. Therefore, by blocking these pathways, ICIs could theoretically provoke an immune response against the fetus [7]. Moreover, as immunoglobulin G antibodies, ICIs can cross the placental barrier with potential transmission from the mother to the developing fetus, potentially resulting in an increased risk of immune-mediated disorders [8,9]. Preclinical studies in murine models and cynomolgus monkeys have shown that the PD-1/PD-L1 blockade in pregnancy disrupts tolerance to the developing fetus. Even if neither malformations nor immunological complications are observed in the offspring [4,5], this dysregulation may result in fetal death [10]. Therefore, using ICIs in pregnant women and women of childbearing potential (in the lack of effective contraception) is not recommended unless the clinical benefit outweighs the potential risk [11]. Nevertheless, some case reports on ICI exposure before, at, or after conception have shown favorable fetal outcomes without developmental abnormalities [12,13,14,15,16,17,18,19,20,21,22,23].Considering the inherent challenges and limitations of pregnancy research, including ethical issues for inclusion in clinical trials, large-scale spontaneous reporting systems represent a privileged source of real-world data to investigate pregnancy-related adverse events. Therefore, to gain further insight into ICI safety in pregnancy, we queried VigiBase®, the World Health Organization’s (WHO) global pharmacovigilance database of safety reports of suspected adverse drug reactions (ADRs) [24]. Herein, we described the largest-to-date series of cases from clinical practice referring to ICI exposure during the peri-pregnancy period and the possible reporting of pregnancy-related outcomes. 4. Discussion

This pharmacovigilance study in VigiBase® provided the largest-to-date series of cases referring to ICI exposure during the peri-pregnancy period and reporting pregnancy-related outcomes. We found no specific patterns of maternal, fetal, or neonatal toxicity. No signal of disproportionate reporting was detected for spontaneous abortion, fetal growth restriction, or prematurity with ICIs, although 104 pregnancy-related outcomes were identified from 56 patients.

Notwithstanding the expanding use of ICIs across multiple indications and settings [1], current knowledge about ICI exposure during the peri-pregnancy period consists of preclinical data and some clinical cases from the scientific literature, with apparently contrasting evidence. On the one hand, studies in animal models with anti-PD-1/PD-L1 agents showed an increased risk of pregnancy loss [4,5,6,7,10], plausibly related to the role played by immune checkpoints in maintaining fetomaternal immune tolerance [2,3]. On the other hand, published case reports of women who were receiving ICI treatment at the time of conception [16,18], or who started ICI treatment during pregnancy [12,13,14,15,17,18,19,21,23], or who became pregnant after the end of ICI treatment [22] overall reported favorable pregnancy outcomes. In contrast, a recent systematic review of case reports also performed a search in the Food and Drug Administration Adverse Event Reporting System (FAERS) and found different pregnancy complications associated with ICIs, including intrauterine growth restriction, spontaneous abortion, premature delivery, and fetal distress syndrome [20]. These findings suggest that published case reports/series could suffer from positive-result bias [40], thus leaving open doubts about ICI safety in pregnancy.In the study cohort, the more common reason for using ICIs was melanoma, which is also the most common malignancy diagnosed during pregnancy [41]. Similar to the published clinical cases [12,13,14,15,17,18,19,21,23], we observed that in most of the safety reports, ICI exposure occurred during pregnancy, whereas no reports indicated ICI treatment at the time of conception. It is noteworthy that three safety reports recorded pregnancy-related outcomes after ICI treatment cessation with a temporal gap ranging from three months to two years. According to current recommendations that advise patients of childbearing age to avoid ICI exposure (unless using effective contraception) during and for at least 5 months after the last dose of ICI treatment [3], the woman reporting an induced abortion 3 months after pembrolizumab discontinuation may still have been exposed to the drug. In terms of efficacy, durable responses with ICIs are becoming increasingly common, suggesting that ICIs could have long-term physiological implications through molecular mechanisms that are still not entirely clear [42,43]. Indeed, the extended duration of their therapeutic effects does not match with their pharmacokinetic half-lives, with an increased risk of late toxicity not only during prolonged active treatments, but also after treatment discontinuation (albeit more sporadically) [43,44,45,46].With regard to pregnancy, immune checkpoint pathways are involved in the establishment and maintenance of maternal immune tolerance to the semi-allogeneic fetus [9]. Therefore, treatment with ICIs could theoretically negatively affect the immune processes underpinning the fetomaternal immune tolerance during pregnancy even months (or years) after the end of ICI treatment. Moreover, besides this direct effect of ICIs on the maternal immune system at the placental barrier, as monoclonal antibodies, ICIs undergo active transport across the placental barrier through the neonatal Fc receptors [47]. These, albeit absent during organogenesis (up to the fourteenth gestational week), increase in the late second and the third trimesters [8,10]. It is noteworthy that individual receptor occupancy might depend on several factors, including tumor burden and genetic polymorphisms, affecting the neonatal Fc receptors [3]. Therefore, because of such an individual variability, long-term toxicities of ICIs during the peri-pregnancy period might occur within a timeframe that cannot be unambiguously defined, thus providing a rationale for the assessment of ICI safety in cases whereby ICI exposure was reported up to two years prior to pregnancy.In preclinical studies, the risk of spontaneous abortion was increased five-fold in models of allogeneic mice pregnancy treated with pharmacological inhibition of PD-L1, but not in syngeneic ones [4], suggesting that fetomaternal immune tolerance and PD-L1 expression at the utero–placental interface could be modulated by the degree of fetal allogeneity [4,7,10]. Therefore, the effects of anti-PD-1/PD-L1 antibodies on the fetus are anticipated to be patient-specific and strongly linked to the paternal antigenic components [43]. Another role whereby the father might be involved is as an oncologic patient treated with ICIs. To date, it is unknown whether paternal treatment with ICIs might affect pregnancy. Nevertheless, it is worth mentioning that in our case series, a few safety reports of paternal ICI exposure were included. Remarkably, in one case, the father was treated with ICIs before conception (although it was unknown how long before ICI treatment was interrupted), whereas in the remnants, the time of exposure was not documented.

Most of the safety reports included in the present study were from healthcare professionals, suggesting a spreading awareness about the potential negative effects of ICI use in pregnant women or those of childbearing age. Moreover, slightly less than half of the safety reports referred to ICI exposure during the peri-pregnancy period without mentioning any type of pregnancy-related outcomes.

This finding might further support the fact that among healthcare professionals, the use of ICIs during the peri-pregnancy period, whenever chosen, remains suspicious and prompts them to spontaneously report their use not in accordance with product labels, even in the absence of pregnancy complications.

Interestingly, we observed that the reporting of safety reports concerning ICI exposure during the peri-pregnancy period peaked in 2019 and then declined over the following two years. This observation might suggest that, despite the widespread awareness mentioned above, the knowledge provided in the last two or three years by clinical cases describing positive pregnancy outcomes could have influenced the attitude of reporters, who, in the absence of pregnancy complications, may have stopped reporting ICI exposure alone. In addition, the decline observed from 2019 onwards could be related to the fact that during the COVID-19 pandemic, in several healthcare systems, a delay of several cancer diagnoses with the postponement of related therapies occurred.

With regard to pregnancy-related outcomes, neither maternal specific pregnancy complications nor patterns of immune-related adverse events were observed, in line with the published single clinical cases [12,13,14,15,16,17,18,19,20,21,22,23]. Noteworthy, one case of maternal hypophysitis occurred, which is a well-described endocrine toxicity with ICIs [48] with potential negative consequences on gonadal function and fertility [9]. Concerning fetal/neonatal outcomes, few safety reports reported successful pregnancy outcomes, and no patterns of major birth defects and no specific immune-related adverse events were found. Strengths and LimitationsAs VigiBase® is the largest spontaneous reporting system collecting safety reports worldwide, disproportionality analyses performed in this database support the generalizability of results. While disproportionality analyses are established approaches for signal detection of rare ADRs, a few studies have specifically addressed pregnancy-related outcomes [25,26,27,28,29,30]. In this setting, which concerns a niche of safety reports, we carried out a rigorous case selection and assessment, accounting for major biases (e.g., confounding by indication and concomitant drugs). Nonetheless, we acknowledged major drawbacks of spontaneous reporting systems, including over- and under-reporting (which also limits the sensitivity of signal detection by disproportionality analysis [31]), partial and missing information, and unavailability of clinical details (in the specific setting of ICIs, concerning, e.g., cancer stage, duration of ICI treatment, patient comorbidities, and exact trimester of ICI exposure). Moreover, VigiBase® does not allow for firmly inferring causality, as safety reports are based on the reporter’s suspicion of a causal relationship between drug use and the onset of adverse events, without information on differential diagnoses. Lastly, follow-up information on children is not collected.

Furthermore, disproportionality analysis per se has some limitations. The ROR is a statistical estimate that does not inform about the real risk of developing a certain ADR, but only indicates an increased risk of reporting that ADR. As the denominator (i.e., the exposed population) is unknown, ROR computation does not inform about the incidence of ADRs. Therefore, disproportionality analysis can only generate hypotheses that eventually need to be further investigated.