Introducing biological features at diagnosis improves the relapse risk stratification in patients with acute promyelocytic leukemia treated with ATRA and chemotherapy

This is the authors' abstract. We don't add key points for this paper.

To the Editor: With the current ATRA and chemotherapeutic regimens, over 80% of patients with de novo acute promyelocytic leukemia (APL) can be cured 1, 2. The PETHEMA group stratified APL patients into three different subgroups (low, intermediate, and high risk), by using upfront WBC count (> or <10 × 109/L) and platelet count (> or <40 × 109/L) 3. A risk-adapted treatment strategy based on this classification has improved clinical outcome, as previously reported 4-6. Although molecular and immunophenotypic prognostic markers have been identified in APL, so far they have not been used for patient stratification and risk-adapted therapy. Here, we tested in multivariate analysis clinico-biological features observed at baseline in a large series of 143 APL patients treated with ATRA and chemotherapy at a single institution observed between March 1993 and December 2010. Our aim was to identify new prognostic factors and improve the risk stratification. Diagnosis was confirmed in all cases by RT-PCR identification of the disease-specific PML/RARA fusion gene. All patients received the same induction schedule with ATRA and idarubicin. Seventy-three patients diagnosed between January 1993 and May 2000 received uniform post-remission treatment based on the AIDA 0493 regimen 7, while 69 patients diagnosed after May 2000 were treated according to the risk-adapted regimen AIDA-2000 with distinct post-induction approaches based on the initial risk stratification as reported 5. In all cases, records were available for complete blood counts, peripheral blood and bone marrow blast infiltration, type of PML/RARA fusion gene, coagulation laboratory parameters, morphologic variant, FLT3-ITD presence, and CD34 expression. Our group and others previously reported of a negative prognostic value of CD56, but in this study we did not consider this antigen for the low rate of positivity in our cohort 8, 9. Molecular tests for PML/RARA monitoring were performed after the third consolidation and then every three months for two years and every six months after the end of maintenance. Molecular relapse was defined as a positive PML/RARA test detected in two successive marrow samples collected at any time after consolidation therapy by RT-PCR and in the absence of morphologically detectable blasts in both marrow and peripheral blood 10. Differentiation syndrome (DS) was defined as definitely present when at least three major signs were recorded according to the criteria of Frankel 11. Screening for FLT3-ITD mutations was carried out as previously reported 12. The Wilcoxon–Mann–Whitney test was performed for comparison of non-parametric series and the Fisher's exact test was used to compare categories. Multivariable Cox proportional hazard regression models were performed and expressed as hazard ratios with 95% confidence intervals (CIs). Values of P < 0.05 were considered statistically significant. Overall survival (OS) was estimated using the Kaplan–Meier method. Clinical features of patients analyzed are shown in Table 1. Several clinical and biological features at baseline were tested in univariate and then in multivariate analysis to identify baseline prognostic factors associated with OS and relapse risk: type of transcript, relapse risk, FLT3-ITD mutation, morphologic subtype, and presence of CD34 maintained an independent prognostic value. Based on the above results, we identified a novel score predicting OS and disease-free survival (DFS) using the following baseline factors: WBC count and platelet count as previously published by the Spanish group, with 0 point for low risk patients, 1 point for intermediate, and 2 points for high risk patients; transcript type, 0 for bcr1/2 and 1 for bcr3; FLT3-ITD mutation, 0 for absence and 1 for the presence of the mutation; morphological aspect, 0 for the classic form and 1 for the variant form; CD34 positivity, 0 if negative and 1 if positive. Accordingly, we constructed a predictive model that stratified patients into three categories: (1) low risk: presenting with 0 or 1 risk factors; (2) intermediate risk: with 2 or 3 risk factors; (3) high risk: with 4–6 risk factors. Based on the above model, 143 patients were stratified as follows: 34 (24%) as low risk, 67 (47%) as intermediate risk, and 42 (29%) as high risk. Of 34 patients originally identified as low risk according to the Sanz score 3, 26 had a 0–1 score and 8 a 2–3 score according to the new stratification. Three of the 26 patients (11.5%) experienced a relapse and the OS was 8.4 years; of the 8 patients reclassified as intermediate risk according to the new score, 50% experienced a relapse and the OS was 6 years (Table 2). Of the 67 patients identified as intermediate risk according to the Sanz score, 27 were reclassified as low risk with the new score: 7% of patients experienced a relapse and the OS was 8.1 years. Thirty-one patients were confirmed as intermediate risk: 26% experienced a relapse and the OS was 7.2 years. Finally, 9 patients had a score 4–6 and were reclassified as high risk according to the new score: 77% of patients experienced a relapse and the OS was 3.7 years. Of the 42 patients classified as high risk according to the Sanz score, 15 were reclassified as intermediate and 27 remained as high risk according to the new score. Three of the 15 patients (20%) reclassified as intermediate experienced a relapse and the OS was 6.5 years, whereas 12 (44%) truly high-risk had a relapse and the OS was only 2.5 years. In addition, we could document a statistical difference in OS between patients with score 4 and score 5–6 (56% vs. 34%, P = 0.02). Overall, the newly proposed scoring system that included biological parameters enabled a refined prognostic stratifications with OS rates of 96%, 87%, and 60% for low, intermediate, and high risk patients, respectively (P = 0.02), and DFS rates of 92.5%, 76%, and 58.4% for low, intermediate, and high risk patients, respectively (P = 0.001). The new stratification also allowed predicting the occurrence of a differentiation syndrome, identifying patient at low (9.4%), intermediate (14.8%), and high risk (22%) of developing the syndrome (P = 0.002). In the present study, we defined clinico-biological features that allow improving the previously established Sanz score with a more refined stratification of patients at baseline, in particular for those included in high-risk category. With regard to the impact of the PML/RARA transcript, several studies 3, 6, 13, 14 have previously reported a trend toward an inferior outcome for BCR3 isoform. On a more mature follow-up, we previously found 15 that patients with BCR3 transcript had a significantly higher relapse rate. A long-term follow-up analysis of our patient series also revealed that FLT3-ITD confers poor prognosis, with a 9 years OS of 96% in the FLT3-ITD−ve cohort compared to 39% in the FLT3-ITD+ve cohort and 9-year CIR rates of 4% and 60% for the FLT3−ve and FLT3+ve cohorts, respectively 12. A joint study reported that morphologic variant subtype had, compared to the classic subtype, higher CIR and significant reduced OS 16. CD34 expression in APL has been associated with hyperleucocytosis, bcr3 isoform, microgranular variant, and CD2 expression 17, 18. In our experience, isolated CD34 expression identified a subgroup with a classic morphology (79%), bcr1 prevalence (53%), higher rate of relapse (37% vs. 13.8% in the negative group, P = 0.002), higher incidence of differentiation syndrome (55% vs. 12%, P = 0.03) and lower OS (60% vs 95%, P = 0.001). Indeed, CD34/CD2 expression identified, as previously reported, a subgroup with variant subtype, bcr3 transcript subtype, high risk according to the risk of relapse, high incidence of differentiation syndrome, lower OS (88% vs. 95%), and a significantly higher rate of relapse (22% vs. 13.8%, P = 0.05) 19. We identified, with this improved stratification, which takes into account also biological parameter, a statistical difference in terms of OS, relapse-free survival, and DFS. In particular, the new stratification is able to identify very high-risk patients (score equal to 6) with very low OS. Further investigations are requested in order to validate this model on independent APL series of patients and to entail the adoption of better-tailored strategies in patients at high risk of relapse. MB designed the study, analyzed data, and wrote the manuscript; MSDP and AG performed phenotypic analysis; DD performed molecular analysis; GC, MM, RL, CM collected data; RF and FLC critically revised the paper and approved final version. Massimo Breccia1*, Maria Stefania De Propris1, Matteo Molica1, Gioia Colafigli1, Clara Minotti1, Daniela Diverio1, Roberto Latagliata1, Anna Guarini1, Francesco Lo-Coco2,3, Robin Foà1 1Division Hematology, Department of Cellular Biotechnologies and Hematology, Sapienza University, Rome, Italy; 2Department of Biomedicine and Prevention, University Tor Vergata, Rome, Italy; 3Laboratory of Neuro-Oncohematology, Santa Lucia Foundation, Rome, Italy