Immune thrombocytopenia after immune checkpoint inhibitor therapy

Immune checkpoint inhibitors (ICIs) are agents that block critical immune checkpoints with potent anti-tumour effects. Although multiple immune checkpoints have been identified, most ICIs in clinical practice target the PD-1 (Programmed cell death protein 1)/PD-L1 (programmed cell death ligand 1) pathway.^1-7 Although generally possessing a favourable safety profile, several unique immune-related adverse events (irAEs) are seen with these agents. Haematologic irAEs have also been described in the literature and are thought to be underrecognized.⁸ The incidence is estimated to be as low as 0·5%⁹ and include aplastic anaemia,^10-13 red cell aplasia,^{14, 15} autoimmune haemolytic anaemia,^{16, 17} thrombotic thrombocytopenic purpura,^{18, 19} and immune-mediated thrombocytopenia.^20-22 Here, we describe our single-institution experience of eight patients who developed immune checkpoint inhibitor-induced immune thrombocytopenia (ir-TCP) and highlight similarities with standard immune thrombocytopenia (ITP).

Results

We retrospectively identified eight patients who developed ir-TCP while being treated with an ICI. Clinical and demographic information is summarized in Table I. Ir-TCP was defined as a nadir platelet count of under 100 000 × 10⁹/l and a greater than 50% decrease from baseline in patients with no other apparent reason for the thrombocytopenia. Two patients merely received checkpoint inhibitors as the only treatment for their malignancies, while 50% (n = 4) received three or more therapies.

All patients received ICIs alone except one patient who received concurrent cytotoxic therapy (carboplatin, etoposide) with ICI. The median time on ICI therapy prior to thrombocytopenia was 17 weeks (range 7–62). All cases had grade 4 thrombocytopenia after receiving ICI except for case 2 (grade 3, nadir platelet count 30). The median baseline platelet count was 146 (range 89–248) while the median nadir platelet count was 17 (range: 2–30). Bone marrow assessments were performed in seven cases without evidence of metastatic cancer or myelodysplasia. Three cases had antiplatelet antibody assessments; two had detectable antiplatelet IgG antibodies (case 2 and case 7). Although HIV, hepatitis B/C, and Helicobacter pylori were not uniformly checked, no case of these or any other infections was identified.

In addition to holding the ICI, corticosteroids were given in all cases as initial treatment for ir-TCP. The most common corticosteroid (50%, n = 4) utilized in our cohort was prednisone at a dose of 1 mg/kg. The response rate (platelet count greater than 50 000 × 10⁹/l) to front-line corticosteroid therapy was 50% (n = 4). The median time to response was 10 days. In four patients, second-line therapy was given consisting of intravenous immunoglobulin (IVIG; 75%, n = 3) and romiplostim (25%, n = 1). Other therapies were as follows: IVIG (three patients) as well as rituximab, fostamatinib, romiplostim, eltrombopag, and allogeneic stem cell transplantation (one patient each). Of the two patients who received a thrombopoietin (TPO) agonist, one responded. Three patients required at least three lines of therapy for ir-TCP with one patient receiving five lines of treatment for ir-TCP, ultimately responding to fostamatinib. Ultimately, all cases except one had resolution of thrombocytopenia with a median time to resolution of 43 days; one other patient with ir-TCP did not have resolution of thrombocytopenia until an allogeneic stem cell transplant which was performed as therapy for the diffuse large B-cell lymphoma (DLBCL).

Only one patient had a rechallenge of ICI therapy (patient 2); thrombocytopenia developed again, which required multiple agents (steroids, IVIG, eltrombopag), and ICI was permanently discontinued after cycle 7 due to the development of autoimmune pancreatitis. Of note, three patients remained with disease control of the underlying malignancy despite lack of resumption of the ICI, with two of those patients having exceptional responses with one complete response (case 5) and one near-complete response with stable disease (case 8) over 12 months without further therapy.

Discussion

Haematologic irAEs, while uncommon, belong to the diverse spectrum of irAEs that may occur with ICI therapy. In particular, clinically relevant thrombocytopenia associated with checkpoint inhibitors is a rare event.^1-4 The mechanism behind irAEs remains an area of active investigation and may have to do with deregulation of tolerant self-reactive T cells and/or cross-reactivity between anti-tumour immune response and normal tissues sharing similar targets.²³ PD-1 plays a critical role in autoimmunity, and blocking PD-1 or its ligands has been shown to accelerate the process of autoimmune diseases.²⁴ Suppression of haematopoiesis directed by T cells has been implicated as a factor in the pathogenesis of immune-related haematologic disorders.²⁵

Typically, ir-TCP develops within the first 12 weeks of therapy initiation,^{9, 20} although it may occur even after cessation of treatment.^26-28 Case 1 is an exceptional case with the patient receiving 38 cycles of nivolumab before experiencing grade 4 ir-TCP. Overall, length of ICI therapy, advanced disease, or extensive prior therapy did not appear to correlate with the severity of the thrombocytopenia, which is consistent with the current literature.^29-31 As with traditional ITP, ir-TCP is a diagnosis of exclusion. In all the cases, there was a clear temporal relationship between ICI administration and the onset of thrombocytopenia. Concomitant medications were reviewed, and drug-induced thrombocytopenia was thought to be highly unlikely, and clinical responses to glucocorticoid therapy and other systemic treatments given for ITP also supported an immune-mediated effect. Bone marrow biopsy was performed in seven patients to exclude a clonal disorder with no cases of therapy-related myeloid neoplasm or marrow involvement by the primary malignancy.

All of our cases received steroids as initial treatment for their ir-TCP following guidelines by the American Society of Clinical Oncology (ASCO),³² which largely mirror the American Society of Hematology guidelines for ITP.³³ At present, ASCO recommends starting 0·5–1 mg/kg/day of prednisone or equivalent for grade 2+ thrombocytopenia and holding ICI therapy until improvement to grade 1 or better; subsequent therapies for ir-TCP are per the management of ITP. A 2016 meta-analysis showed that response rates for front-line corticosteroid therapy in ITP ranged from 59% to 79%,³⁴ and 50% of the patients in our series responded to initial corticosteroid therapy, with the other 50% requiring further therapy for the thrombocytopenia. All our cases except one had resolution of thrombocytopenia (median time to resolution 43 days); another patient with persistent thrombocytopenia in spite of multiple lines of therapy ultimately proceeded to allogeneic stem cell transplantation for management of the underlying disease, and still another responded to fostamatinib given as fifth-line therapy.

There are a few considerations particular to ir-TCP compared to de novo ITP including ICI rechallenge and appropriate platelet thresholds for intervention. A recent retrospective analysis of 24 079 irAE cases revealed 25·4% experienced irAE with ICI rechallenge,³⁵ and the only case of ICI rechallenge in our series resulted in recurrent ir-TCP as well as the development of a second irAE (autoimmune pancreatitis). On the other hand, responses to ICI are often durable, and there is a significant concern of disease progression once therapy is interrupted. Furthermore, there is also increasing use of ICI in the adjuvant setting such as in advanced-stage melanoma where suspension of therapy may compromise the probability of cure. In our cohort, none of the cases had clinical significant bleeding requiring intervention. Therefore, the platelet threshold at which intervention is indicated should be tailored individually as thrombocytopenia in advanced cancer patients can be problematic.

While fairly low platelet counts are typically tolerable in ITP, severe thrombocytopenia in cancer patients can complicate decisions regarding further therapy and anticoagulation, both of which are frequent concerns in this population. One final provocative point is the observation of a potential correlation between irAE and improved clinical outcomes;³⁶ it has been posited that ir-TCP may be associated with improved clinical outcomes.³⁷ One of our patients (case 5) did indeed have an exceptional response to immunotherapy and remains in complete response from his metastatic colorectal cancer 18 months after the last cycle of pembrolizumab. In conclusion, ir-TCP is an uncommon entity which requires prompt recognition and individualized management in order to avoid interruption of therapy. Novel therapies for ITP, in particular TPO mimetics, deserve further study in this setting as steroid sparing agents given the efficacy and excellent safety profile.