Intensive Chemotherapy Is Associated With Poor Overall Survival in Autoimmune Disease-associated Myeloid Malignancies.

Between 10% and 20% of cases of myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML) are associated with prior exposure to cytotoxic chemotherapy or ionizing radiation.[1] Such therapy-related myeloid neoplasms have also been associated with treatment for nonmalignant conditions, notably autoimmune diseases (AID) such as inflammatory bowel disease (IBD) and rheumatoid arthritis (RA).[2] Furthermore, the chronic inflammatory state associated with AID has been associated with an increased risk of myeloid malignancies, independent of the treatments used.[3] A recent study of over 40,000 patients with AID showed a statistically significant 7-fold increase in therapy-related myeloid neoplasms in patients treated with azathioprine, with smaller, nonsignificant risks attributable to cyclophosphamide and mitoxantrone, but not anti-tumor necrosis factor therapy (in contrast with lymphoma).[4,5] While therapy-related myeloid neoplasms are generally associated with poor prognosis, there are limited published outcome data in the AID setting. In a single-center experience of 23 patients with AID-associated AML, most exhibited intermediate (43%) or favorable risk (26%) cytogenetics and intensive treatment (including allogeneic transplantation) was associated with a median overall survival (OS) of 68 months.[6] Of note, this cohort was relatively young (median 59 years) and the 83% of cases were female, both of which associate with more favorable prognosis in AML/MDS. This contrasts with smaller case series that associate AID-AML/MDS with poor-risk cytogenetics and short OS.[7-9]

We report the characteristics and treatment outcomes of 19 consecutive patients presenting to University College London Hospital with AML/MDS on a background of treatment for AID between August 2011 and March 2018 (Table 1). The median age at presentation with AML/MDS was 62 years (interquartile range [IQR] 50–70 years) with a secondary peak at age 19 years (n = 3). There was an equal sex distribution (M = 10, F = 9). The most common AIDs were IBD (ulcerative colitis n = 6, Crohn disease n = 1), myasthenia gravis (n = 3), and RA (n = 3). The median time from diagnosis of AID to myeloid malignancy was 10 years (IQR 76–251 months). The commonest immunosuppressive agents used were azathioprine (n = 10), methotrexate (n = 5), sulfasalazine (n = 5), and cyclophosphamide (n = 2). Three cases (patients 5, 11, and 15) had no exposure to immunosuppressive therapy associated with myeloid malignancies. Despite the presence of underlying AID, the baseline performance status of the majority of patients was good (18/19 ECOG 0), although some patients had significant comorbidities (Table 1).

Table 1

Summary of Patient Characteristics

Thirteen patients (68%) presented with features consistent with AML. Six patients (32%) presented with MDS (n = 3) or chronic myelo-monocytic leukemia (CMML) (n = 3). Patients exposed to azathioprine presented with both AML (n = 5) and MDS (n = 5), whereas those exposed to sulfasalazine, methotrexate, and cyclophosphamide all presented with AML. The majority of patients presented with modest peripheral leukocyte counts (median 5.3 × 109/L, IQR 2.4–11.0 × 109/L, max 150 × 109/L) and relatively preserved hematopoiesis (median hemoglobin 96 g/L, IQR 88–114 g/L; platelets 88 × 109/L, IQR 26–139 × 109/L). Median bone marrow blast percentage was 25% (IQR 12–60%). One patient presenting with AML had good-risk cytogenetics and remains in long-term complete remission (CR) after intensive chemotherapy. One patient with unopposed NPM1 mutation died of respiratory failure during induction chemotherapy. The majority (53%) of patients exhibited poor-risk cytogenetics (n = 10). Six of seven patients tested by next-generation sequencing harbored poor-risk mutations.[10] Overall, the majority of patients (68%, n = 13) harbored an adverse cytogenetic or molecular marker, with 21% intermediate-risk (n = 4, including 2 with no cytogenetic/molecular results) and 11% (n = 2) good-risk.

Ten patients were treated with intensive induction chemotherapy, including both favorable risk patients (median age 52 years, AML [n = 9]). Eight patients received nonintensive treatment with azacitidine (median age 67, range 58–84, MDS/CMML n = 6) and 1 patient declined treatment. Of the patients undergoing an intention-to-treat with intensive induction chemotherapy, only 3 (30%) received more than 1 cycle of first-line treatment, with 50% of these patients dying within 40 days of diagnosis, principally from sepsis and/or respiratory failure. Three patients treated intensively received re-induction chemotherapy at relapse, none of whom achieved CR. One patient treated with intensive chemotherapy responded sufficiently to proceed to allogeneic bone marrow transplantation. By contrast, the median number of cycles of azacitidine administered at time of analysis was 6 (range 2–15). Two patients who progressed on azacitidine underwent salvage treatment: patient 9 successfully achieved a second CR with intensive chemotherapy; patient 13 was treated with venetoclax, but died of refractory AML. Overall, CR was achieved in a total of 6 patients (3/10 intensive, 3/8 azacitidine).

Median OS for all patients was surprisingly poor at 167 days (Fig. 1A). Unexpectedly, AID-AML/MDS patients treated with intensive chemotherapy demonstrated a significant survival disadvantage over those treated nonintensively with azacitidine (median OS 79 vs 263 days, hazard ratio 2.911, 95% confidence interval 0.87–9.73, P = 0.045) (Fig. 1B).

Figure 1

Analysis of overall survival outcomes. (A) Overall survival probability of all patients with AID associated myeloid neoplasms (median = 167 days, n = 19). (B) Overall survival probability comparing induction treatment with intensive multiagent chemotherapy (n = 8) to the demethylating agent azacitidine (n = 8) in cytogenetically intermediate/poor-risk AML/MDS secondary to AID (79 vs 263 days, HR 2.911, 95% CI 0.87–9.73, P = 0.045). AID = autoimmune disease, AML = acute myeloid leukemia, CI = confidence interval, HR = hazard ratio, MDS = myelodysplastic syndrome.

Our experience suggests that AID-AML/MDS is associated with adverse prognostic factors and a median OS shorter than expected in secondary-AML/MDS. Patients with intermediate/poor-risk AID-AML/MDS treated with intensive chemotherapy have markedly high induction mortality, low rates of CR and dismal OS. By contrast, the demethylating agent azacitidine was not associated with significantly reduced OS in AID patients and should be considered as a less toxic alternative to intensive chemotherapy in this cohort. This is the second-largest published cohort of AID-AML/MDS treatment outcomes, with results that are markedly inferior to those from the larger Philadelphia series.[6] This disparity likely relates to biases in retrospective series, such as differences in age or gender. Further studies are required to identify AID-AML/MDS patients who benefit from intensive induction chemotherapy and investigate novel curative treatment strategies, in particular whether azacitidine can provide a safe and effective bridge to allogeneic transplantation. Lastly, as high-dose chemotherapy is increasingly used to treat advanced AID, our experience suggests that clinicians should exercise caution and proactively report adverse outcomes.