Prognostic Factors of Infections and Effect of Primary Anti-Infectious Prophylaxis in MDS Patients Treated with Azacitidine (AZA): A Prospective Study

Background: Hypomethylating agents, especially AZA, have become the reference first line treatment of high-risk MDS. Myelosupression, although less important than with chemotherapy, is however observed, leading to potentially life threatening infections. A retrospective study found unfavorable (unfav) cytogenetics and low platelet counts to be predictive factors of infections in high risk MDS and AML patients (pts) receiving AZA (Merkel and al, Am j Hemat 2012). However, prognostic factors of infections, and whether infection prophylaxis would be useful in this situation, has not been prospectively evaluated.

Methods: Between June 2011 and March 2013, 120 high-risk MDS pts were included in a randomized phase II trial seeking the most promising drug association with AZA by comparison with AZA alone in higher risk MDS (including AML with 20 to 30% marrow blasts and CMML with > 10% marrow blasts) (NCT01342692). Pts received AZA (75mg/m²/dx7d every 4 weeks) alone (N=40), with Valproic acid (N=40) or with Lenalidomide (N=40) (10mg/dx14d every 4 weeks). G-CSF was not used. Infectious events (IE) (diagnosed as such by the treating physician), hospitalizations for sepsis and pts receiving antimicrobial prophylaxis were reported at each cycle. Predictive factors of the occurrence of IE were analyzed.

Results: 75 (62.5%) pts developed 259 IE, including 61 requiring hospitalization in 46 pts (61.3% of infected patients). The number of IE and of infected patients were similar in the 3 study arms. 39 pts died during the study, 12 of them because of infection, none of whom had responded to AZA (4 progressions, 4 failures and 4 deaths before evaluation). IE were more common during the first two cycles of therapy, with 86 (31.3%), 52 (23.5%) 45 (18.9%), 26 (15%), 15(19.2%) and 24 (19.7%) IE during cycles 1, 2, 3, 4, 5, and 6, respectively. Fever of unknown origin (FUO) (39.6%) and pneumonia (28.8%) were the most common type of infections followed by ENT (9.9%), urinary tract (8.1%), skin (5.4%), dental (4.5%) and intra-abdominal (3.6%). 6,3% were bacteriemia. Among the 26 microbiologically documented IE, 13 were CG+ (4 staph aureus, 4 enteroccus species, 4 coag neg staph and 1 other), 9 were BG- (6 E Coli, 1 pseudomonas and 2 others) and 3 were viral (HSV1, influenza B, Hepatitis E) and only one patient had documented invasive fungal infection (asp fumigatus).

Overall, 23 (19%), 22 (18%), 10(8%) pts received bacterial (Levofloxacine), fungal (posaconazole) and viral (Valaciclovir) prophylaxis resp. Predictive factors of IE were unfav karyotype (79.5% infections vs. 50.8% in pts with fav or int karyotype; p=0.005) and platelets (PLT) < 20 G/L (92.3% infections vs. 58.9% for platelets > 20 G/L; p=0.03). In multivariate analysis, only unfav karyotype was predictive of IE (p=0.01). Other baseline parameters (including ANC, IPSS, age, sex, Hb level, and BM blast %) and bacterial, fungal or viral prophylaxis had no significant predictive value on the occurrence of IE. In multivariate analysis, predictive factors of pulmonary infection were anemia at baseline (p=0.04) and unfav karyotype (p<0.001), while prophylaxis had no significant impact. Infected pts had significantly more hospitalizations and deaths than non-infected pts (p<0.0001 and p=0.028 resp.). In multivariate analysis, unfav karyotype (p<0.001) and PLT <20 G/L (p=0.05) were significantly predictive of hospitalization for infection, while baseline Hb <10g/dL (p=0.02), and unfav karyotype (p=0.03) were predictors of fatal infection.

Conclusion: 62.5% of the 120 pts developed infections during AZA treatment, mainly during the first 2 cycles, and 10% of the pts died from infection. Only one invasive fungal infection was documented. Unfav karyotype was strongly predictive of IE, hospitalization for infection and fatal infections. Other significant predictive factors were baseline anemia for pulmonary infection and fatal infection, and thrombocytopenia for hospitalization for infection, while ANC was not a significant factor. Moreover, prophylaxis was not associated with a decrease of IE in our study, but the small number of pts who received it precludes any conclusion.