Real‐world efficacy and safety of immunochemotherapy in cold agglutinin‐mediated autoimmune haemolytic anaemia

In cold agglutinin (CA)-mediated autoimmune haemolytic anaemia (AIHA), autoantibodies bind to the ‘I’ antigen on red blood cells (RBCs) at temperatures below 37°C.1 They are typically monoclonal and of immunoglobulin (Ig)M class. The clinical picture is marked by complement-mediated haemolytic anaemia and cold-induced peripheral symptoms (CIPS).2 Cold agglutinin syndrome (CAS) refers to cases secondary to infections, autoimmune diseases or overt haematological malignancies.3 However, most patients with CA-mediated AIHA have cold agglutinin disease (CAD), with evidence of a clonal B-cell disorder in blood and/or bone marrow.4, 5 Although symptoms can sometimes be managed through cold avoidance and supportive care, the majority of CA-mediated AIHA patients require systemic treatment.2 In CAS, treatment of the underlying condition is the preferred approach.3 Several treatment options exist for CAD, each with advantages and limitations, but no consensus on the optimal strategy. Rituximab (R-) monotherapy remains the most established therapy targeting the underlying B-cell clone. However, response rates are ±50% and median response duration less is than 1 year.6, 7 Immunochemotherapy (ICT) is more effective, as demonstrated in a prospective trial of 45 CAD patients treated with R-bendamustine, achieving response rates of 70%–80% and a 5-year sustained remission of 77%.2, 8 However, it is associated with increased toxicity. Other ICT regimens, particularly R-cyclophosphamide-based combinations, are being applied as well, based on experience in low-grade B-cell malignancies like Waldenström macroglobulinaemia (WM).9 However, evidence on their efficacy in CAD and CAS remains limited. More recently, newer targeted treatments have emerged, including complement inhibitors, anti-CD38 therapy and Bruton's tyrosine kinase inhibitors, which may require long-term maintenance therapy and are more expensive.10-12 Moreover, complement inhibitors do not relieve CIPS, as the underlying antibody-producing clone is not targeted. Thus, it is important to optimally position each option in the expanding treatment landscape for CA-mediated AIHA. A multicentre retrospective study was performed to assess the efficacy and safety of ICT in CAD and CAS with real-world data. All consecutive CAD and CAS patients from 20 Dutch centres were included if they had a direct antiglobulin test strongly positive for complement and received at least one ICT cycle between 2000 and 2024 (Figure S1). Eligible ICT regimens included R-bendamustine- and R-cyclophosphamide-based therapies, such as dexamethasone-R-cyclophosphamide (DRC), R-cyclophosphamide-prednisone (R-CP) and R-cyclophosphamide-vincristine-prednisone (R-CVP). Since the distinction between CAD and CAS can be difficult in clinical practice and is not uniformly applied, both patients with CAD as well as CAS secondary to a low-grade B-cell lymphoma were considered eligible. Patients with mixed AIHA were excluded, as well as patients with CAS, if secondary to an infection or autoimmune disease. This study was conducted in accordance with the Declaration of Helsinki and reviewed by the Medical Research Ethics Committee of Amsterdam University Medical Center. Follow-up duration for response assessment and early toxicities was defined as the time from ICT initiation to the start of the next treatment (for CAD or underlying disease), death or the end of data capture, whichever occurred first. Treatment response was assessed by hemoglobin (Hb) levels, transfusions, CIPS, total immunoglobulin M (IgM) and monoclonal (M-) protein. Additionally, a combined Hb/CIPS response was established, and modified CAD response criteria, adapted from Berentsen et al., were applied when data were available.8 Response criteria are specified in Table S1. Adverse events (AEs) were graded according to Common Terminology Criteria for Adverse Events v5.0.13 Event-free survival (EFS) was defined as the time from ICT initiation to death of any cause or subsequent treatment (for AIHA or underlying disease), whichever occurred first; patients were censored at the last data capture. For the occurrence of second malignancies, monitoring continued until the last contact or death. A total of 34 patients were included (17 R-cyclophosphamide and 17 R-bendamustine); 18 (53%) were female, and the mean age at diagnosis was 62 ± 10 years (Table 1). In all 34 patients, a monoclonal gammopathy was detected in peripheral blood prior to the start of ICT, of which 32 (94%) were of IgM kappa isotype. In 28 patients (82%), bone marrow assessment showed evidence of a clonal B-cell population, of which 13 (46%) were classified as an overt B-cell malignancy (see Supplementary Information). Seventeen patients received R-cyclophosphamide-based treatment. Median follow-up duration for response evaluation was 27 months (interquartile range; IQR 14–39) (Figure 1A). Of the 15 anaemic patients (Hb <12 g/dL) at start, 6 (40%) reached a complete and 5 (33%) a partial Hb response, while 4 (27%) had no response. Haemolysis resolved in 5 (45%) of the 11 responders. Median time to first Hb response was 1.4 months (IQR 1.1–6.4, Figure S2A). At baseline, six patients (40%) had a recent transfusion (in the preceding 3 months). Four of those achieved a transfusion-free interval of >3 months. At baseline, CIPS were reported in 12 patients (71%). Of these, seven (58%) achieved clinical improvement (with complete resolution in three; 25%), while four (33%) had no improvement and one was unknown. The combined Hb/CIPS response was evaluable in 16 patients (94%); 9 (56%) achieved any response on both parameters, if applicable. The CAD response criteria by Berentsen et al. were assessable in 14 patients (82%).8 One patient (7%) achieved an unconfirmed complete response (CR), five (36%) a partial response (PR) and eight non-response (NR) (57%). Median EFS was 28 months with an estimated 5-year EFS of 26% (Figure S3A). Follow-up ended due to new treatment for AIHA-related symptoms in 11 patients, new treatment for underlying disease in one, and death due to suspected high-grade transformation 10 months after ICT initiation in another. Grade ≥3 AEs occurred in seven patients (41%); these included three grade 4 AEs and seven grade 3 AEs (Table S2). Five patients (29%) had toxicity-related dose adjustments or schedule alterations. Second malignancies during long-term monitoring (median 65 months; range 10–211) occurred in four patients (24%), comprising myelodysplastic syndrome (MDS; n = 1; in a patient that received multiple lines of ICT), basal cell carcinoma (n = 1), colon carcinoma (n = 1) and high-grade transformation of WM (n = 1). Seventeen patients were treated with R-bendamustine. Median follow-up duration for response was 29 months (IQR 14–60) (Figure 1B). Of the 10 anaemic patients (Hb <12 g/dL) at start, 8 (80%) reached a complete Hb response and 1 a partial Hb response, while 1 had no response. Haemolysis resolved in seven of the nine (78%) responders. Median time to first Hb response was 6.8 months (IQR 3.3–13.2, Figure S2B). At baseline, two patients (12%) had recent transfusion history. One of those achieved a transfusion-free interval of >3 months. At baseline, CIPS were reported in 12 patients (71%). Of these, 10 (83%) achieved clinical improvement (with complete resolution in five; 42%) and 2 patients (17%) experienced no improvement. The combined Hb/CIPS response was evaluable in all 17 patients; 15 patients (88%) achieved any response on both parameters (if applicable). The CAD response criteria by Berentsen et al. were assessable in 15 patients (88%).8 Five patients (33%) achieved an unconfirmed CR, seven (47%) a PR and three NR (20%). Median EFS was not reached, and the estimated 5-year EFS was 83% (Figure S3B). Follow-up ended due to the start of another treatment for AIHA-related symptoms in one patient and death due to progressive anaemia, infectious complications and possible progression of underlying LPL in another. No grade 4 AEs occurred, and five patients (29%) experienced an AE of grade 3 (Table S2). Nine patients (53%) had toxicity-related dose or schedule alterations. In one patient, a second malignancy occurred during long-term monitoring (mean 40 months, range 8–94), comprising a melanoma. Of the 17 patients who initially received R-cyclophosphamide, 5 subsequently received R-bendamustine and 1 received a second R-cyclophosphamide regimen, followed by R-bendamustine. These treatment lines are discussed in the Supplementary Information. In summary, the combined Hb/CIPS responses in our cohort were 56% and 88% for R-cyclophosphamide-based regimens and R-bendamustine respectively. While this retrospective study is neither designed nor suitable for a head-to-head comparison, R-bendamustine seemed associated with better response rates and a trend towards longer duration of response, despite this group having received more prior treatments and exhibiting a longer time to response. Alternatively, the limited response rate in patients receiving R-cyclophosphamide may reflect the requirement for deeper clonal responses to elicit clinical benefit, similar to what has been observed in other monoclonal gammopathies of clinical significance.14 Of note, although based on small numbers, Hb responses in transfused patients were often absent or short-lived in both groups. In general, the overall Hb/CIPS response of 88% in our study aligns with the overall response of 78% in the prospective R-bendamustine trial.2, 8 However, different criteria were applied due to missing data, thereby compromising direct comparison. Furthermore, our estimated 5-year EFS of 83%, although similar, cannot be directly compared to the 77% 5-year response duration as reported by Berentsen et al., as we used time to new treatment as a proxy for response duration due to the lack of well-established, retrospectively applicable relapse criteria. Both regimens came with a high-grade toxicity rate of 30%–40%. Notably, toxicity-related treatment alterations occurred in half of the R-bendamustine–treated patients, emphasizing the clinical importance of low-grade AEs. The development of second malignancies or transformation cannot be attributed to ICT with any certainty, but is possibly treatment related in one case of high-grade transformation of WM and one emergence of MDS after three ICT lines. Limitations of this study relate to its retrospective design and the rarity of the disease, including reporting bias, missing data and early loss of follow-up. Interpretation of CIPS outcomes is complicated by its multifactorial nature (due to potentially co-existing cryoglobulins, which were not routinely tested), seasonal variability, subjectivity and the absence of a validated symptom score.15 However, our data support the notion that ICT can improve CIPS, as opposed to complement inhibitors. While comparing outcomes between CAD and CAS would be relevant, this distinction is not always evident and is inconsistently applied in practice, limiting feasibility in a retrospective setting. We present the first real-world experience of ICT in CA-mediated AIHA in a relatively large Dutch cohort. Our findings support ICT, particularly R-bendamustine, as a valid fixed-duration treatment option offering durable symptom relief for both haemolytic anaemia and CIPS. Disadvantages remain the relatively long time to response and unfavourable toxicity profile. FVMM designed the research, collected and analysed the data, and wrote the manuscript. ALB designed the research, collected and analysed the data, performed statistical analysis and wrote the manuscript. JMIV designed the study, supervised the data analysis and reviewed the manuscript. SJBM supervised the data analysis and reviewed the manuscript. BILW supervised statistical analyses and reviewed the manuscript. MJ, DE, LtB, AB, JD, RvdG, KdH, ABUM, CMPWM, LN, RO, JFMP, JRe, JRo, LS and FMvdV provided clinical data and reviewed the manuscript. All authors provided final approval of the manuscript. The authors thank all additional Dutch haematologists who contributed patient data to this series, including Gert-Jan Timmers, Tim de Waal, Cees Schaar, Wim Terpstra and Henriëtte Levenga. FVMM was supported by Sanquin Blood Supply foundation (PPO-SHS-2022-22.13/L2687); ALB was supported by Mr. R. de Jong through the Amsterdam UMC foundation. JMIV has received the following as institutional honoraria: research support from Beigene and AbbVie/Genmab; advisory board/consultancy fees from Sanofi and Janssen; and speaker fees from BMS, Sanofi, Beigene, Novartis and Amgen. For the remaining authors, no relevant conflicts of interest were declared. This study was conducted in accordance with the Declaration of Helsinki, and reviewed by the Medical Research Ethics Committee of Amsterdam University Medical Center. All participants provided written informed consent or, if deceased, no opt-out decision was recorded during their lifetime. Study data are available from the corresponding author upon reasonable request. Data S1. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.