Azacytidine Failure Revisited: An Appraisal Based on Real Life Data from the MDS Registry of the Hellenic Myelodysplastic Syndrome Study Group (HMDS)

Azacytidine (AZA) is the mainstay of treatment for high-risk MDS, but both primary and secondary AZA failure confers a grave prognosis. Given the limited availability and efficacy of subsequent treatment options the exact definition of AZA failure and the decision to discontinue AZA is of particular importance in everyday clinical practice. However, the current definition of AZA failure is rather obscure and often lumps together pathobiologically diverse disease states, as reflected by the highly heterogeneous outcome of patients who fail AZA (post-HMA model, Nazha A et al, 2016).

We retrospectively analyzed data from 323 non selected MDS, MDS/MPN and low blast count AML (LBC-AML) patients enrolled to the MDS AZA registry from July 2004 to May 2017. The current analysis included only patients who received AZA as 1^st line treatment. The 7 and 5-2-2 days, i.e. a 2-day break, regimens were used. Treatment response was evaluated using the IWG 2006 criteria. Survival analysis was performed using a Kaplan-Meier estimate and Cox's proportional hazards model. Overall survival (OS) was defined as the time from AZA initiation to last follow up or death from any cause.

Table 1 lists general patient information. After a median follow up of 41.7 months the median OS from AZA initiation for the whole cohort was 13.5 (95% CI: 11.8-15.2) months. The French Prognostic Scoring System (FPSS, Itzykson, 2011) was the better survival discriminator compared to IPSS, WPSS and IPSS-R (Fig. 1) and still functioned better if the analysis was limited to high-risk MDS and LBC-AML patients (n=232). In multivariate analysis, FPSS (p<0.001), best response to AZA (CR/PR vs all others, p<0.001) and disease subtype (MDS/MPN vs others, p=0.01) were all independent predictors of OS. Median time to AZA discontinuation was 7.4 months (95% CI: 6.4-8.4) with 42(13%) patients still receiving AZA at the time of analysis. Progressive disease and toxicity were the most common causes of AZA discontinuation, conferring identical median OS (5.5 vs 5.2 months, respectively, p=0.9), whereas most patients were treated with supportive care and only 9 patients proceeded to allo-SCT (table 1).

Median OS from AZA failure, defined as no response after 4-6 cycles with AZA intolerance, progressive disease or death while on treatment, and loss of response, was 5.6 (95% CI: 4.9-6.3) months. The post-HMA model provided the better prediction of OS after AZA failure (7.2 vs 4.6 months for low vs high risk respectively, p=0.02), while FPSS functioned better than the rest systems (Fig 2). In multivariate analysis only the post-HMA model retained its independent predictive value.

Interestingly, patients who continued AZA despite loss of response had a significantly better OS compared to those who stopped AZA immediately at loss of response (8 vs 4.6 months, p=0.04, Fig 3). According to the post-HMA model 16 patients who discontinued AZA were high and 7 low risk, while in those who continued AZA 15 were high and 15 low risk. Though the decision to stop or not AZA when the response is lost according to the IWG criteria is obviously biased, patients who continued AZA had still significantly longer OS (p=0.04) even when the parameter of stopping or not AZA was adjusted for disease subtype and the post-HMA model, whereas both patient groups received comparable therapies after AZA discontinuation.

In summary, in our large patient dataset we confirmed the superior efficacy of FPSS and post-HMA models in predicting OS at AZA start and failure, respectively. However, our data also indicate that in real life settings AZA failure is rather ill-defined, and physicians' perceptions of the cause and timing of AZA discontinuation differ widely. Failure due to intractable toxicity and disease progression confers very poor outcome, while patients who lose response to AZA appear to fare better and may benefit from the continuation of AZA despite poor prognostic features, highlighting on one hand the diversity of AZA resistance mechanisms and, on the other, the limitations of the applicability of prognostic models and response criteria in everyday practice.

View largeDownload slide

View largeDownload slide

Close modal