AUTHORS
Alessandro Laganà, MD, Filippo Branzanti2, Emilia Scalzulli, MD3, Maria Laura Bisegna, MD4, Chiara Sartor, MD5, Ida Carmosino6, Costanza Andriola, MD7, Deborah Kasmi, MD8, Alessandra Dedola9, Giovanni Iaquinta, PhD10, Maurizio Martelli, MD11, Francesca Palandri, MD, PhD12, Massimo Breccia, MD13
AFFILIATIONS
1 Sapienza Università di Roma, Roma, Lazio, Italy
2 IRCCS Azienda Ospedaliero-Universitaria di Bologna, Bologna, Emilia-Romagna, Italy
3 Sapienza Università di Roma, Roma, Lazio, Italy
4 Sapienza Università di Roma, Roma, Lazio, Italy
5 IRCCS Azienda Ospedaliero-Universitaria di Bologna, Bologna, Emilia-Romagna, Italy
6 Sapienza Università di Roma, Roma, Lazio, Italy
7 Sapienza Università di Roma, Roma, Lazio, Italy
8 Sapienza Università di Roma, Roma, Lazio, Italy
9 IRCCS Azienda Ospedaliero-Universitaria di Bologna, Bologna, Emilia-Romagna, Italy, Università di Bologna, Bologna, Emilia-Romagna, Italy
10 San Camillo-Forlanini Hospital, Roma, Lazio, Italy
11 Sapienza Università di Roma, Roma, Lazio, Italy
12 IRCCS Azienda Ospedaliero-Universitaria di Bologna, Bologna, Emilia-Romagna, Italy
13 Sapienza Università di Roma, Roma, Lazio, Italy
ABSTRACT
BACKGROUND: In Myelofibrosis (MF), conventional prognostic scores are primarily designed for risk stratification at diagnosis and are not designed for patients (pts) undergoing ruxolitinib (RUX) therapy. Currently, there is a lack of baseline tools specifically tailored to predict outcomes upon JAK-inhibitor (JAKi) initiation, as the RR6 score provides post-6-month-treatment guidance.
AIMS: To identify baseline clinical predictors of overall survival (OS) in RUX-treated MF pts and develop an easily and broadly applicable prognostic model at RUX initiation.
METHODS: We analyzed two independent, real-world cohorts of JAKi-naïve, RUX-treated MF pts: derivation cohort from Sapienza University (n=232) and external validation cohort from the University of Bologna (n=125). RUX-Eligibility was based on national indications: PMF or SMF; DIPSS≥int-1; palpable spleen ≥5 cm below costal margin; PLTs ≥50×109/L. To ensure cohort homogeneity and minimize confounding due to the distinct prognostic profile of early-PMF, these pts were excluded from both cohorts. For OS analysis, pts undergoing allo-HCT were censored at transplant date. Restricted cubic splines were utilized to identify optimal prognostic cut-offs for continuous variables. Multivariate Cox proportional hazards model for OS was constructed by including univariate-screened covariates with a p<0.10 along with established, clinically relevant prognostic factors regardless of their univariate significance. Model performance was assessed using C-index, AIC, and reclassification metrics (NRI/IDI).
RESULTS: In the derivation cohort, median age at RUX start was 66.2 years; 122 were males (52.6%) and 98 (42.2%) had a PMF. Median Hb was 11.0 g/dl, median PLT count 292.0×109/L and 44.4% had ≥1 peripheral blast (PB). G3 bone marrow fibrosis (BMF) was detected in 53.9%. DIPSS risk distribution was: 54.7% int-1, 35.8% int-2, and 9.5% high. After a median follow-up of 71.6 months, the overall IWG-MRT-spleen response (SR) rate was 48.3%. At 6 months of RUX, RR6 distribution was: 32.9% low, 42.1% intermediate, and 25.0% high risk. Disease progression to acute myeloid leukemia (AML) occurred in 42 pts (18.1%); 89 deaths (38.4%) were recorded, with an estimated median OS of 78.0 months.
Multivariate analysis identified the following baseline factors as independent predictors of shortened OS: age ≥68 years (HR 3.41; 95%CI 2.03–5.72), G3 BMF (HR 2.42; 95% CI 1.42–4.10), anemia (Male<11 g/dL, Female<10 g/dL) (HR 2.17; 95%CI 1.29–3.66), PLTs < 150×109/L (HR 2.18; 95%CI 1.31–3.65), PB 1–2% (HR 1.75; 95%CI 1.03–2.97) and ≥3% (HR 4.16; 95%CI 2.19–7.90). The R-PSS was developed by assigning weighted points to the identified predictors and pts were classified into four distinct risk categories: Low (0-2), Int-1 (3-5) (HR 7.34, 95%CI 2.51–21.49, p<0.001), Int-2 (6-8) (HR 18.09, 95%CI 6.26–52.23, p<0.001) and High (≥9) (HR 60.76, 95%CI 20.57–179.46, p<0.001). Internal validation via bootstrapping (1000 samples) confirmed a p=0.001 for all risk tiers. Median OS was not reached for the Low-risk group, and was 81.8, 47.5 and 26.4 months for Int-1, Int-2 and high tiers, respectively (p<0.001). The R-PSS demonstrated a significantly higher C-index (0.800) compared to both baseline DIPSS (0.739; p=0.007) and RR6 score (0.705; p<0.001) and achieved the lowest AIC (720.9 vs 758.4 for RR6 and 774.2 for DIPSS). Moreover, at 60 months, the R-PSS significantly improved the accuracy of risk assignment over DIPSS by 51.1% (NRI 0.511, p=0.005) and over RR6 by 40.6% (NRI 0.406, p=0.012), as well as model's overall predictive power by 17.9% over the DIPSS (IDI 0.179, p=0.004) and by 22.2% over the RR6 (IDI 0.222, p<0.001).
Interestingly, R-PSS showed a good accuracy in predicting the absence of 6-month and overall SR, with an AUC of 0.703 and 0.748, respectively. R-PSS was also able to stratify PFS risk with a C-index of 0.789. In the external validation cohort, the R-PSS demonstrated a high discriminative capacity for both OS (C-index 0.718) and PFS (C-index 0.71), with median OS not reached for the low-risk group, 88.7, 47.5 and 26.0 months for Int-1, Int-2, and High tiers, respectively (p<0.001).
CONCLUSION: RUX remains an effective standard of care for MF. In this context, the clinically based and easily applicable R-PSS may help in identifying patients at high risk of treatment failure, supporting early therapeutic decision-making and optimal patient selection for novel strategies and/or allo-HCT referral.
DISCLOSURES
Alessandro Laganà, MD: No relevant conflicts of interest to declare.