Computational intelligence identifies alkaline phosphatase (ALP), alpha-fetoprotein (AFP), and hemoglobin levels as most predictive survival factors for hepatocellular carcinoma

Liver cancer kills approximately 800 thousand people annually worldwide, and its most common subtype is hepatocellular carcinoma (HCC), which usually affects people with cirrhosis. Predicting survival of patients with HCC remains an important challenge, especially because technologies needed for this scope are not available in all hospitals. In this context, machine learning applied to medical records can be a fast, low-cost tool to predict survival and detect the most predictive features from health records. In this study, we analyzed medical data of 165 patients with HCC: we employed computational intelligence to predict their survival, and to detect the most relevant clinical factors able to discriminate survived from deceased cases. Afterwards, we compared our data mining results with those obtained through statistical tests and scientific literature findings. Our analysis revealed that blood levels of alkaline-phosphatase (ALP), alpha-fetoprotein (AFP), and hemoglobin are the most effective prognostic factors in this dataset. We found literature supporting association of these three factors with hepatoma, even though only AFP has been used in a prognostic index. Our results suggest that ALP and hemoglobin can be candidates for future HCC prognostic indexes, and that physicians could focus on ALP, AFP, and hemoglobin when studying HCC records.

International Prognostic Indexes (IPI, RIPI, NCCN-IPI) for Predicting Overall Survival in Diffuse Large B-Cell Lymphoma Can be Improved By Adding Serum Albumin As Another Predictive Factor. Retrospective Study of the Mexican Study Group in Lymphoma from a Database of 855 Patients

Background: There are well-established prognostic factors for predicting the overall survival (OS) of patients involved by diffuse large B cell lymphoma (DLBCL). These prognostic factors are age, performance status (PS; ECOG score), high serum lactate dehydrogenase level (LDH), advanced disease (Ann Arbor stage: III, IV) and extranodal sites involved (EN). The patients are distributed in different risk groups according to the score obtained and as established by each research group who developed each international index. However, these prognostic indexes do not explore new markers and just readjust the five known variables since 1993 when The International Non-Hodgkin's Lymphoma Prognostic Factors Project was developed. Serum albumin (SA) has been shown to be a prognostic biomarker in DLBCL prior to R-CHOP treatment (Ngo et al. Leuk Lymph 2008). In another study of patients older than 80 years receiving R-miniCHOP treatment, SA ≤3.5 g/dl was the only factor with a significant effect on OS on (Peyrade et al Lancet Oncol 2011). Studies of non-Hodgkin's lymphoma reported that SA <3.0 g/dl was one of the factors predicting early death from aggressive disease treated with ACVBP (Dumontet et al. Br J Haematol 2002). Another study found that SA <3.7 g/dl is an independent prognostic indicator in DLBCL patients treated with R-CHOP (Dalia et al. Ann Hematol 2014). Aims. The present study evaluates the behavior of IPI, RIPI and NCCN-IPI risk groups obtained from a database of 855 patients involved by DLBCL and explores a cut-off for SA and the addition of it as part of those risk groups. Methods. Seven databases were obtained from different public and private institutions in the country with patients diagnosed with DLBCL. Creating a unique database with 855 patients. Only 811 patients had serum albumin levels in their record (94.8%). Statistical methods: First, we decided to obtain the SA cut-off through the Received Operating Curve because any study has evaluated this (including sensibility, specificity, likelihood ratio positive and negative). Thus, we evaluated the influence of SA in OS. For that, OS was assed using Kaplan-Meier method (KM) and compared between obtained groups using log rank-test. Subsequently, we add SA score to the different risk prognostic index groups and were assed using KM again. Hazard ratio was evaluated using cox-regression model (CRM). Outcomes. From 2008 to 2016 were included from seven different databases in one, 855 patients of different institutions (publics, academics and privates) in different States. Demographic characteristics: Female,51.46%; Median age was 63 y/o (range: 18-96), ≥ 60, 54.8%; PS, ECOG ≥2,31.1%; EN ≥ 1, 32%; LDH > normal, 52.74%. Type of regimen: Rituximab + anthracycline-based regimen= 75%, anthracycline-based regimen=10.5%, rituximab + CVP= 7%, Other treatments=4.5%, No treatment= 3%. All of them were included as intent-to-analyses. The classical variables were evaluated using the CRM, for IPI and RIPI; age and LDH were evaluated according to the proposed cut-off or ratio respectively in the original paper for NCCN-IPI, confirming that all were significant as predictor factors (p <0.0001). The Area under the curve for SA in relation to OS was 0.70 (CI95% = 0.67 to 0.73, p <0.0001) and the cut-off point was <3.2g / dL. However, we observed that SA had two cut-off points with different predictive value that perfectly differentiated two groups. These points were for low group: 3.2 g / dL to 2.5 g / dL (PPV=68%, NPV= 68.1%) and for very low group: ≤2.4 g/dL (PPV=80% , NPV=60.5%). Then, we evaluated its influence in OS, normal SA was 64% ±0.02 vs low group 34%±0.03 (HR=2.6, CI95%=2to3.5) vs. Very low group 16%±0.04 (HR=4.5, CI95%=3to7). For the addition of SA to IPI, RIPI and NCCN-IPI, we gave a 1 point to Low and 2 points to very low; Then, these points were added to the risk scores of the different IPI, which increased the final scores or not. In Table 1, the percentages of OS of each risk group can be compared before and after adding the points according to the SA score. Where adding SA significantly improves the distribution of risk groups. See table 1. Conclusion: SA is an important predictive factor of OS in patients involved by DLBCL. Two risk groups can be observed according to the SA level. The addition of SA to the prognostic indexes improves the distribution of patients and the OS percentage at 5 years of follow-up. Figure. Figure. Disclosures Gomez-Almaguer: AbbVie: Consultancy; Novartis: Consultancy.

The Engineered MIPI (e-MIPI), a Candidate Data-Mining Based Mantle Cell Lymphoma Prognostic Index Developed from the Dataset of the Fondazione Italiana Linfomi (FIL) MCL0208 Phase III Trial

Background and aims The amount of clinical and biological data stored within clinical trials is growing exponentially. Data warehousing (DW) is useful for systematic global evaluation of information collected in trials: the highly translational FIL(Fondazione Italiana Linfomi)-MCL0208 trial has been used to test DW to improve data quality and to discover putative associations [Zaccaria, ASH 17]. In this study we developed an engineered prognostic model, focusing on easily accessible clinical variables. For this purpose, we exploited hierarchical clustering with the aim of seeking hidden patterns of interest in large datasets. Hence, these tools allowed to develop a novel prognostic model: the engineered MIPI index (e-MIPI). Herein we present the first results, on baseline clinical characteristics:clustering analysis and definition of a signature of predictive variablesconstruction of the e-MIPI to detect patients' risk of relapsecomparison with known prognostic indexes for MCLvalidation of the signature on independent subset of patients. Methods Data were retrieved from electronic case report forms of the phase III, multicenter FIL-MCL0208 trial (NCT02354313) for younger MCL patients [Cortelazzo, EHA 15]. The study enrolled 300 subjects, with median followup of 51 months. In this work we employed baseline clinical data and May '18 as survival outcomes cut-off. For the present analysis, we started from 32 baseline features: 7 were not eligible due to number of missing values (MVs ≥40). Features with <15 MVs were imputed by median of observations. Secondly, 18 not binary variables were dichotomized, to be compared to the 7 binary ones: normal vs abnormal range or lower vs higher than a recognized cut-off value. Patients were thus split in 2 subsets, training (n=185) and validation (n=115): for the training set, only patients with no MVs were chosen. Clustering analysis was performed to discriminate different groups of patients. Thus, we applied a recursive feature reduction, according to regression modeling, to extrapolate a restricted signature predictive of both progression free survival (PFS) and overall survival (OS). Survival analyses were done according to e-MIPI classes via both multivariate Cox and Kaplan-Maier modeling. Therefore, the e-MIPI classification was compared to known prognostic models [Hoster, Blood 08]. Finally, the signature was tested on the validation set: if any variable of the e-MIPI was missing (MVs=36, 29 and 15 for albumin - alb, Ki67 and flowcytometric peripheral blood invasion - flowpb) data mining (K-nn) technique was employed for imputation. Clustering and statistical analyses were implemented via MATLAB© and SPSS©. Results Training set: the clustering analysis allowed to define 3 groups of subjects: C1 (n=71), C2 (n=77) and C3 (n=37), showing significantly different PFS and OS. Thus, the e-MIPI index was modeled based on a signature of 9 significant features (fig 1): histologic bone marrow infiltration (bminf), flowpb, Ki67, B symptoms, platelets (plts), ldh, white blood cells (wbc), hemoglobin (hb) and alb levels. The re-clustering of the training set according to the e-MIPI confirmed the original patients clustering with 83% of accuracy. Figure 2A depicts the PFS curves stratified for the e-MIPI: C1, C2 and C3 groups have been renamed as low (L), intermediate (I) and high (H) e-MIPI risk classes, respectively. Each comparison reached the statistical significancy: I vs L, p=0.010; H vs I, p=0.023, outperforming in our series both the MIPI-St (H vs I risk, p=0.801) and MIPI-Bio (I vs L risk, p=0.665, fig. 2B) classifications. Validation set: the e-MIPI allowed to discriminate 3 groups of subjects C1 (n=32), C2 (n=59) and C3 (n=24). Actually, the e-MIPI on the validation set (fig. 2C) confirmed the results of the training set, overall improving the MIPI-St stratification (H vs I, p=0.059 ⇒ p=0.049), even if without reaching the statistical significancy on the I vs L comparison (p=0.24 ⇒ p=0.15), due to the limited number of events in this series. Discussion e-Mipi is a new first prognostic index derived from hierarchical clustering. Our results indicate that this approach might allow to model engineered prognostic indexes based on comprehensive analysis of large datasets. Even if promising, it needs validation through its application to independent series of MCL patients. Additional efforts aiming at integrating biological variables in the model are ongoing. Disclosures Gaidano: Amgen: Consultancy, Honoraria; Morphosys: Honoraria; Janssen: Consultancy, Honoraria; Gilead: Consultancy, Honoraria; Roche: Consultancy, Honoraria; AbbVie: Consultancy, Honoraria. Ladetto:Celgene: Honoraria; Sandoz: Honoraria; Jannsen: Honoraria; Roche: Honoraria; Abbvie: Honoraria; Acerta: Honoraria.