Cyclin-dependent kinase 4/6 inhibitors and interstitial lung disease in the FDA adverse event reporting system: a pharmacovigilance assessment

The cyclin-dependent kinase (CDK)4/6 inhibitors – palbociclib, abemaciclib, and ribociclib – are now approved both in the United States and Europe for women with hormone receptor-positive, human epidermal growth factor receptor 2-negative advanced breast cancer, based on positive results of several large pivotal phase III randomized controlled trials [1‐3].

From a safety standpoint, CDK4/6 inhibitors are well-tolerated agents, with similar safety profile, although some differences exist in the pattern and frequency of toxicities, which might influence the choice of a given medication. The most common side effect for palbociclib and ribociclib is neutropenia, whereas gastrointestinal toxicity is associated especially with abemaciclib (showing less selectivity for CDK4, which plays a critical role in hematopoietic stem cell differentiation) [4]. Among rare adverse events, higher frequency of QT prolongation emerged for ribociclib, whereas transaminases increase was recorded with ribociclib and abemaciclib resulting in regulatory warnings [5].

Pulmonary toxicity was reported for several anticancer drugs, and over 1300 medications, procedures, or substances are considered to cause respiratory impairment (www.​pneumotox.​com), although the pulmonary liability is often recognized after marketing approval. The term “interstitial lung diseases” (ILDs) poses a challenging clinical diagnosis as it refers to heterogeneous disorders with remarkably different clinical pathophysiology, histories, and prognoses, such as idiopathic pulmonary fibrosis, sarcoidosis, connective tissue disease associated with ILDs, and hypersensitivity pneumonitis. Drug-induced ILD incidence rates vary between 4.1 and 12.4 cases/million/year, with anticancer agents, anti-rheumatic drugs, amiodarone, and antibiotics being the most common causes [6].

Although data on pulmonary toxicity with CDK4/6 inhibitors are scant, recent case reports have described the potential occurrence of pneumonitis, including fatal cases [7]. Moreover, in the Food and Drug Administration (FDA) label for abemaciclib, it is reported that there was 1 death due to pneumonitis in the MONARCH 3 trial, and 2 deaths due to pneumonitis in the MONARCH 2 trial. In the MONALEESA-2 trial, there were two deaths secondary to acute respiratory failure in patients receiving ribociclib plus letrozole. In May 2019, the Japanese Ministry of Health released a warning as 14 patients in Japan developed pulmonary toxicity likely due to abemaciclib exposure (nearly 2000 exposed), and three of them died. Recently, on September 13th, 2019, the FDA issued a warning about rare but severe lung inflammation with CDK4/6 inhibitors, based on cases of ILD and pneumonitis identified in the manufacturers’ completed and ongoing clinical trials and the post-marketing safety databases. Although rare (1–3% of patients had ILD/pneumonitis of any grade), there were serious cases (less than 1% had fatal outcome), and some patients had at least one risk factor (https://​www.​fda.​gov/​drugs/​drug-safety-and-availability/​fda-warns-about-rare-severe-lung-inflammation-ibrance-kisqali-and-verzenio-breast-cancer).

Post-marketing monitoring is therefore crucial to timely characterize ILD and to target preventive strategies for diagnosis and management [8]. In this context, international spontaneous reporting systems, through collection of millions of worldwide reports, represent a primary source of data for safety assessment of recently marketed drugs receiving fast track designation and priority review, which deserve rigorous post-marketing monitoring [9, 10]. In particular, the FDA Adverse Event Reporting System (FAERS) is the largest publicly available pharmacovigilance database particularly suitable to detect rare adverse events, which may escape detection and/or reporting from randomized controlled trials.

On these grounds, the aim of this real-world post-marketing study is to comprehensively characterize spontaneous reports of ILD with CDK4/6 inhibitors submitted to FAERS and investigate whether pulmonary toxicity actually occurs with all CDK4/6 inhibitors.

In the past few years, the therapeutic scenario of hormone receptor-positive, epidermal growth factor receptor 2-negative, advanced/metastatic breast cancer has been profoundly changed by the clinical availability of CDK4/6 inhibitors. These agents differ in terms of tolerability (gastrointestinal, liver, and bone marrow toxicities), with relevant impact on safe prescribing by oncologists [1‐5].

To the best of our knowledge, this is the first large-scale post-marketing safety study investigating the reporting of ILD with CDK4/6 inhibitors. This pharmacovigilance assessment, stemming from recent serious case reports of pneumonitis, characterizes the occurrence of ILD through multiple disproportionality approaches combined with a case-by-case evaluation, a complementary approach only rarely carried out in the literature [25]. Although this pulmonary toxicity appears to be rare, oncologists should be aware that ILD does occur with CDK4/6 inhibitors even in patients without apparent drug-related risk factors, namely concomitant drugs known for their ILD risk.

Overall, we found a consistent increased reporting of ILD with CDK4/6 inhibitors as a class, which was consistent across the various sensitivity analyses accounting for major biases and confounders, especially for abemaciclib. ILD is key feature of Japanese population, as indicated by several post-marketing analyses documenting a higher-than-expected reporting of ILD in Japan, especially for anticancer drugs, possibly due to genetic susceptibility, market penetration, and Japan regulatory system [20‐22]. Notably, removal of Japan reports did not have substantial impact on disproportionality for all CDK4/6 inhibitors, and a signal was even unmasked for palbociclib and ribociclib, thus suggesting that a drug-related component is likely to exist. The large reporting of ILD in elderly population (as compared to other events) should be also further explored as a potential host-related risk factor.

Although actual incidence and prevalence cannot be derived from spontaneous reporting, the fact that CDK4/6 inhibitors were reported in 161 cases over the past few months (1.2% out of total 13,450 ILD cases in the past 4 years) suggests that the estimated risk is not so rare as stated in the relevant summary of product characteristics (SPC); this is also in line with a recent polled safety analysis of MONARCH 2 and 3 trials, describing ILD as an infrequent toxicity [26]. Remarkably, the proportion of ILD vs other events exponentially increased in 2009, even before the FDA warning; this trend can be potentially interpreted as a consequence of the increasing uptake of CDK4/6 inhibitors worldwide, especially in progressively less selected populations. Therefore, we fully endorse the FDA warning. The relevant sections in the US and EU labels differ and provide imprecise mention of the relevant ILD profile. Palbociclib was the only agent with a warning in both US and EU SPCs, whereas abemaciclib and ribociclib received a warning in the EU and US SPCs, respectively. We believe that the SPCs should be harmonized, also by adding relevant data in the specific section on side effects, where only data from clinical trials are provided.

When looking at the pneumotox.​com (last access July 09th, 2020), a standard reference for ILD, the statement “pneumonitis/ILD, including acute and severe form (may occasion an ARDS picture),” is reported for palbociclib and abemaciclib, whereas for ribociclib, a general reference to the FDA warning on CDK4/6 inhibitors as a class is mentioned. We found a highly probable/probable causality assessment in the majority of cases, thus corroborating a true drug-related association. Moreover, positive dechallenge supports the reversibility of ILD (reported to occur at therapeutic doses); these clinical elements, together with the observed relatively rapid time to onset (median latency of 63 days for CDK4/6 inhibitors as a class, in line with literature data on anticancer drugs) [27], strengthen the importance of early recognition of signs/symptoms suggestive of ILD (e.g., dyspnea), especially at the beginning of treatment, for appropriate management including drug discontinuation. In MONARCH 2 and 3 trials, only rarely were dose reductions and/or discontinuation required, whereas ¼ of patients were treated with steroids and/or antibiotics [26].

Abemaciclib emerged with a consistent strong disproportionality (i.e., very high ROR values, mirroring the association with bleomycin). Although we recognize that comparative safety studies through pharmacovigilance databases are debated [28], there are no expected major distortions to the data that can potentially explain the observed difference in disproportionalities, and the three agents are broadly comparable in terms of therapeutic uses, while differing from scheduling of administration (continuous for abemaciclib rather than 3-week cycles, with exposure increasing less than proportionally with an increasing dose) and pharmacodynamic standpoints. Abemaciclib possesses unique pharmacological properties as compared to palbociclib and ribociclib; in vitro data have shown that it is the most potent inhibitor of CDK4 and 6, with lower potency against CDK1, CDK7, and CDK9 and also showed direct inhibitory activities against other kinases including glycogen synthase kinase 3α/β and calmodulin-dependent protein kinase II α/β/γ. Moreover, abemaciclib targets the dual specificity tyrosine phosphorylation-regulated kinase as well as homeodomain-interacting protein kinase [29‐31]. Taken together, these data should prompt mechanistic studies to investigate whether CDK4/6 or other kinases are potentially involved in ILD pathophysiology. Moreover, the potential relationship between physiochemical/pharmacokinetic features (e.g., lung distribution) and ILD occurrence should be explored; of note, abemaciclib possesses higher lipophilicity vs other CDK4/6 inhibitors. These pharmacological differences might also partially explain the observed differences in the latency (50 days for abemaciclib vs 253 days for ribociclib), although the low number of cases does not allow firm conclusions. From a pharmacokinetic perspective, the three agents share some main features, including metabolism mediated by CYP3A4 (with abemaciclib producing active and abundant metabolites), potentially resulting in drug–drug interactions, and biliary clearance as the main elimination pathway [32].

Because of inherent limitations [10‐12, 28], including no certainty on causation, incomplete/missing information, and potential codification errors, these findings raise the hypothesis that a drug-specific susceptibility exists for ILD, and ask per se for additional analytical studies, such as population-based investigation, to confirm the signal before any regulatory action other than information or harmonization of SPCs can be envisioned [30]. In particular, this study cannot be used to quantify ILD risk and provide risk ranking mainly because: (a) under-reporting and the lack of data on population exposure do not actually allow calculation of incidence rate; (b) the diagnosis depends on a number of criteria, including radiological assessment and the exclusion of other causes such as prior radiotherapy, which cannot be obtained with absolute certainty. Moreover, contribution of additional drugs with underlying (but unknown) pulmonary toxicity cannot be excluded, as well as residual confounders. Nonetheless, our real-world analysis has important strengths [33, 34], including a large-scale comprehensive pharmacovigilance assessment, encompassing disproportionality approaches (major confounders ruled out, including notoriety bias) and case-by-case evaluation (WHO and Bradford Hill criteria), which consistently support the notion of a real causal association. While the role of ethnicity deserves further investigation, there are no reasons to support the existence of confounding by indication or channeling bias (i.e., preferential prescription towards more severe patients with risk factors for ILD) and, if present, it should apply to all CDK4/6 inhibitors.

The mechanisms of ILD are unknown, and someone may even consider ILD with CDK4/6 inhibitors a paradoxical event since the anti-proliferative mode of action of these agents can be theoretically useful in counteracting lung fibrosis. However, a recent in vivo study in a mice model of bleomycin-induced lung fibrosis found that, after 14 days, palbociclib significantly decreased collagen deposition in the lung after bleomycin treatment but did not ameliorate lung function. Importantly, palbociclib augmented inflammatory cell recruitment (including macrophages and T cells) in the bronchoalveolar lavage fluid [35]. Mechanistically, the elevated levels of inflammatory cell could be a consequence of the palbociclib-induced cell cycle arrest with relevant cellular senescence, a phenomenon called “senescence associated secretory phenotype.” This supports the biological basis of ILD with CDK4/6 inhibitors and relevant inflammatory-mediated mechanism, which is also compatible with the observed early latency.

In summary, CDK4/6 inhibitors should be added to the evolving list of drug-induced ILD, an emerging and challenging differential diagnosis especially in the current COVID-19 era. The strong signal for abemaciclib, consistent across various approaches, warrants further investigation through population-based studies, to establish actual event rates and identify risk factors that might lead to proper risk management, including the contributing role of age, ethnicity and pharmacological properties.

These findings support the notion that CDK4/6 inhibitors have different safety profile, and ILD should be considered within the spectrum of rare toxicities together with liver injury and QT prolongation. This should be taken into account for safe prescribing in the individual patient, considering the underlying risk factors and therapeutic alternatives.

The relatively recent marketing approval of CDK4/6 inhibitors, together with their promising place in therapy and relevant expected increasing utilization in clinical practice, suggest the importance of monitoring patients at the beginning of therapy for early signs of lung toxicity and strengthen the role of spontaneous reporting systems as a crucial source to monitor in the real-world the occurrence of serious, rare, and unpredictable toxicities. Detailed reporting can provide valuable opportunities to identify safety signals and ultimately protect patients’ safety.