Blood-Based HALP-LDH Score Predicts Survival in Lymphoma Patients

A simple calculation from four routine blood values, combined with a standard enzyme measurement, may help clinicians identify which patients with diffuse large B-cell lymphoma face the greatest risk after first-line treatment. A retrospective study from Zhejiang Provincial People’s Hospital in China reports that a composite score built from the hemoglobin–albumin–lymphocyte–platelet (HALP) index and serum lactate dehydrogenase (LDH) outperformed each marker alone and the widely used International Prognostic Index (IPI) in predicting three-year overall survival. The work, published in Frontiers in Physiology, analyzed 112 newly diagnosed patients who all received R-CHOP immunochemotherapy between January 2014 and October 2021.

The HALP score is a composite measure calculated by multiplying hemoglobin in grams per liter, albumin in grams per liter, and the absolute lymphocyte count in cells per liter, then dividing by the platelet count. Each component captures a different dimension of the patient’s condition: hemoglobin reflects tissue oxygenation and chronic inflammatory burden, albumin signals nutritional status and exerts anti-inflammatory effects, lymphocytes indicate cellular immune competence, and platelets participate in both clotting and tumor microenvironment signaling. Because all four values come from standard hematology and biochemistry panels, the score requires no specialized equipment or additional cost beyond routine testing.

LDH, by contrast, is a tumor-oriented marker. As the enzyme catalyzing the interconversion of pyruvate and lactate, it sits at the core of aerobic glycolysis, the metabolic shift by which rapidly dividing cancer cells consume glucose. Serum LDH levels correlate with tumor burden, proliferation rate, and metabolic activity, and elevated values are already embedded in the NCCN-IPI as an indicator of high-risk disease. The researchers reasoned that pairing a host-state measure with a tumor-burden measure would capture complementary prognostic information that neither provides alone.

To test this, the team divided the 112 patients into a high-HALP group (score above 24.92, n = 65) and a low-HALP group (score at or below 24.92, n = 47), using a cutoff selected by receiver operating characteristic analysis and maximized log-rank statistics. Patients in the low-HALP group carried significantly more adverse features: 76.60 percent had Ann Arbor stage III–IV disease versus 49.23 percent of the high-HALP group, 57.45 percent had IPI scores of 3–5 versus 29.23 percent, 55.32 percent had LDH above 250 U/L versus 30.77 percent, and a larger share had two or more extranodal sites of involvement. All of these differences reached statistical significance at P < 0.05. Age, sex, ECOG performance status, and Hans classification (GCB versus non-GCB subtype) did not differ significantly between the groups.

Treatment response also tracked with the score. The objective response rate—complete or partial remission at the end of treatment—was 87.69 percent in the high-HALP group but only 65.96 percent in the low-HALP group. Conversely, 34.04 percent of low-HALP patients had stable or progressive disease, compared with 12.31 percent of high-HALP patients. The study assessed responses according to the Lugano classification, and the difference in response rates was statistically significant.

Survival outcomes diverged sharply. Over a median follow-up of 34.5 months (range 2–125 months), during which 50 patients died, the high-HALP group showed significantly better progression-free survival (χ² = 7.171, P = 0.007) and overall survival (χ² = 10.200, P = 0.001). Median PFS was 61.0 months in the high-HALP group versus 14.0 months in the low-HALP group (HR = 1.974, 95% CI 1.196–3.170, P = 0.007). Median OS was not reached in the high-HALP group but was 32.0 months in the low-HALP group (HR = 2.523, 95% CI 1.430–4.451, P = 0.001). At 20 months, PFS was 66.86 percent in the high-HALP group versus 43.32 percent in the low-HALP group; at 40 months the corresponding figures were 57.50 percent and 30.18 percent.

Multivariable Cox regression, treating HALP and LDH as continuous variables, identified both as independent prognostic factors for overall survival. Each unit increase in HALP carried a hazard ratio of 0.983 (95% CI 0.968–0.999, P = 0.037), confirming a protective effect, while each unit increase in LDH carried a hazard ratio of 1.001 (95% CI 1.000–1.001, P = 0.007). However, when the same analysis was applied to progression-free survival, HALP lost independent significance (HR = 0.993, P = 0.259), whereas LDH remained significant (HR = 1.001, P = 0.006). The authors attribute this dissociation to the biology of the endpoints: overall survival is strongly shaped by host factors such as nutritional adequacy, inflammatory balance, and immune competence, which modulate treatment tolerance and late mortality, while progression-free survival is governed more by tumor-intrinsic proliferation kinetics that HALP does not directly capture.

The composite HALP+LDH model, built by fitting a Cox regression with both variables and deriving a risk score formula (Risk score = −0.4651 × HALP + 0.1337 × LDH), achieved an area under the ROC curve of 0.717 for three-year overall survival. This exceeded the AUC of HALP alone (0.663), LDH alone (0.690), and the IPI score (0.687), with the improvement over HALP alone reaching statistical significance (DeLong P < 0.01). A parallel HALP+IPI model matched the same AUC of 0.717. The combined model showed 75 percent sensitivity and 68 percent specificity, compared with 58 percent sensitivity and 71 percent specificity for HALP alone and 72 percent sensitivity and 63 percent specificity for LDH alone. Using a risk-score cutoff of 0.24065, patients classified as high-risk had a median OS of 35.0 months (95% CI 16.272–53.728) versus not reached in the low-risk group (log-rank χ² = 16.131, P < 0.001). In a stratified Cox model adjusted for ECOG status, the HALP+LDH risk stratification remained independently associated with OS (HR = 3.268, 95% CI 1.591–6.716, P = 0.001).

Internal validation with bootstrap resampling corrected for optimism confirmed moderate discrimination. For overall survival, the bootstrap-corrected C-index was 0.663 for the HALP+LDH model, 0.641 for the IPI-alone model, and 0.679 for the HALP+IPI model. For progression-free survival, the corresponding values were 0.613, 0.588, and 0.605. Adding HALP to IPI produced a modest improvement in OS discrimination but limited gain for PFS. The authors note that these C-index values, while statistically meaningful, remain moderate in absolute terms and reflect the single-center, modest sample size of the cohort.

The study has clear limitations that the authors acknowledge. It was a single-center retrospective analysis of 112 patients, which introduces potential selection bias and limits external validity; the findings require confirmation in larger, multi-center prospective cohorts. Molecular markers such as MYC/BCL2 double expression, TP53 mutation status, and dynamic ctDNA clearance were not incorporated into the model. The cohort was also restricted to patients without concurrent malignancy, autoimmune disease, active infection, or significant organ impairment, which may not reflect the full spectrum of clinical practice. Nevertheless, the authors argue that the HALP+LDH model occupies a practical niche: it uses only routinely available laboratory values, requires no imaging or molecular assays, and can be deployed in resource-limited settings where PET-based metabolic tumor volume, radiomics, or circulating tumor DNA monitoring remain inaccessible. As a complementary tool rather than a replacement for advanced modalities, the composite score offers a low-cost method for early identification of high-risk DLBCL patients and may support individualized treatment decisions alongside conventional indices such as the IPI.

Subject of Research: Prognostic value of a HALP-LDH composite blood-based score in newly diagnosed diffuse large B-cell lymphoma patients treated with R-CHOP

Article Title: Prognostic analysis of HALP -LDH composite model in newly diagnosed diffuse large B-cell lymphoma patients

Article References: Yan, Y., Li, K., Yu, Q., & Qiu, L. (2026). Prognostic analysis of HALP -LDH composite model in newly diagnosed diffuse large B-cell lymphoma patients. Frontiers in Physiology, 17, Article 1877689. https://doi.org/10.3389/fphys.2026.1877689

Image Credits: AI Generated

DOI: 10.3389/fphys.2026.1877689

Keywords: diffuse large B-cell lymphoma, HALP score, lactate dehydrogenase, R-CHOP, prognosis, overall survival, progression-free survival, International Prognostic Index, biomarker, risk stratification, retrospective cohort study, hematologic malignancy