Serum albumin, but not gnri, is an independent prognostic factor for survival after first allogeneic SCT: A large-scale multicenter analysis

Abstract Background: While we previously reported that the geriatric nutritional risk index (GNRI) is a strong independent prognostic factor for overall survival (OS) and non-relapse mortality (NRM) after second allogeneic stem cell transplantation (SCT2) (Kaito et. al. Ann Haematol 2020, Sadaga et. al. Blood 2023 (Suppl 1)), it remains unclear whether the GNRI and other nutritional indicators are effective prognostic factors for first SCT (SCT1). This study aimed to clarify the prognostic value of pre-transplant nutritional indices, including GNRI, body mass index (BMI), and serum albumin (Alb), in patients undergoing SCT1 to stratify the post-transplant survival. Methods: We retrospectively analyzed 3,178 patients with hematopoietic malignancies who underwent SCT1 between 2004 and 2020 at transplant centers of the Kanto Study Group for Cell Therapy (KSGCT), using the KSGCT database. The primary endpoint was 5-year OS, and the secondary endpoints were 5-year NRM and cumulative incidence of relapse (CIR). We defined acute leukemia and malignant lymphoma (ML) in the first or second complete remission, CML in the first or second chronic phase, and MDS with refractory anemia as standard risk, and other malignant diseases as high risk. Results: A total of 3178 patients who underwent SCT1 were evaluated. The median age was 51 years (range: 16–72). The underlying diseases included AML (n=1494, 47.0%), ALL/LBL (n=642, 20.2%), MDS (n=523, 16.5%), CML (n=110, 3.5%), ML (n=224, 7.1%), and other (n=185, 5.8%). The median follow-up period for survivors was 1959 days (range: 51–5702). The cohort was randomly divided into exploratory (n=1588) and validation (n=1590) cohorts with no significant differences in baseline patient characteristics. In the exploratory cohort, 5-year OS, NRM, and CIR after SCT1 were 51.1%, 21.4%, 35.0%, respectively. To evaluate prognostic value of nutritional indicators including GNRI, Alb, and BMI, receiver operating characteristic (ROC) curve analyses were conducted. The area under the curve (AUC) for OS was 0.597 for GNRI, 0.612 for Alb, and 0.504 for BMI. Alb had significantly better predictive accuracy than GNRI and BMI (p<0.001). The concordance index (C-index) was 0.572 for GNRI and 0.581 for Alb, again supporting the superiority of Alb. Therefore, we selected Alb as the nutritional indicator of OS after SCT1. Through univariate and multivariate analyses on the dependent variables for OS, age at SCT1 (≥55 years) (hazard ratio [HR] 1.500, p<0.001), high C-reactive protein (CRP) (≥0.22 mg/dL; HR 1.428, p<0.001), high-risk disease status at SCT1 (HR 1.924, p<0.001), SCT1 from related haploidentical donor (HR 1.769, p<0.001), high HCT-CI score (≥3) (HR 1.384, p<0.001), poor performance status at SCT1 (≥2) (HR 1.703, p<0.001), low Alb at SCT1 (<3.9 g/dL) (HR 1.381, p<0.001), and transplantation before 2016 (HR 1.236, p=0.005) were independently associated with poor OS. To predict OS after SCT1, we developed a scoring system consisted of seven factors: age ≥55 years (+1), high CRP level (≥0.22 mg/dL, +1), high-risk disease status (+2), high HCT-CI score (≥3, +1), poor performance status (≥2, +1), and low Alb (≤3.9 g/dL, +1). Patients were stratified into three risk groups based on total scores: low risk (score 0-1, n=505), intermediate risk (score 2-4, n=808), and high risk (score ≥5, n=227). OS at 5 years was 71.6%, 46.4%, and 21.2% for patients with low, intermediate, and high risk group, respectively (p<0.001). NRM (13.4%, 23.7%, and 32.8% at 5 years, p<0.001) and CIR (23.8%, 36.6%, and 49.3% at 5 years, p<0.001) were also significantly different among the groups. Notably, all differences between three risk groups for OS, NRM, and CIR were statistically significant using the Bonferroni correction. These findings were further confirmed in the validation cohort, which also demonstrated significant stratification in SCT outcomes among three groups (5y-OS; 73.0%, 46.9%, and 22.9%, p<0.001: 5y-NRM, 14.3%, 22.7%, and 30.4%, p<0.001: 5y-CIR, 20.8%, 36.3%, and 50.4%, p<0.001). Conclusion: In contrast to our findings in SCT2, this study demonstrates that serum Alb, but not GNRI, is the most reliable nutritional prognostic marker for SCT1. Along with low Alb, older age, high CRP, high-risk disease, high HCT-CI, and poor PS were also independent predictors of poorer survival. Incorporating Alb and these factors into a scoring system provided clear and statistically robust risk stratification after SCT1.