Simple Blood Ratio Reveals Hidden Disease Burden in Aggressive Leukemias

Two routine laboratory measurements—serum lactate dehydrogenase and serum albumin—can be divided to yield a single number that researchers have increasingly labeled a nutrition–inflammation biomarker. But what does that number actually measure? A new study from Ganzhou People’s Hospital in Jiangxi, China, published in Frontiers in Nutrition, tackles this question head-on in one of the most genetically aggressive forms of blood cancer: TP53-mutated acute myeloid leukemia (AML) and myelodysplastic syndromes (MDS). The findings suggest that the lactate dehydrogenase-to-albumin ratio, or LAR, is far less a mirror of nutritional status than a window onto the sheer burden of malignant disease coursing through the marrow.

TP53 is often called the guardian of the genome, and when it is mutated in myeloid neoplasms, the prognosis is grim. These cancers resist standard therapies, including newer venetoclax-based combinations, and survival remains limited even with allogeneic stem-cell transplantation. Yet outcomes are not uniform: variant allele frequency, blast count, complex karyotype, and treatment all shape the clinical course. Against this backdrop, the research team led by Xiaoqin Xin and Zongqi Wang of Southern Medical University asked whether a simple, routinely available ratio could complement these established prognostic factors in an exploratory way—and, crucially, what biology the ratio actually reflects.

The study enrolled 116 patients diagnosed between 2016 and April 2026, restricting the primary analysis to 114 adults. The cohort was split almost evenly between MDS (52.6 percent) and AML (47.4 percent), with a median age of 65 years. Nearly three-quarters carried a complex karyotype, and co-occurring myeloid mutations—most frequently DNMT3A and TET2—were present in two-thirds of patients. Median LAR was 7.01 units per gram, and the team modeled it as a continuous, log-transformed variable so that every effect estimate corresponds to a doubling of the ratio, avoiding the arbitrary thresholds that plague many biomarker studies.

The first major result concerns what LAR correlates with. Using Spearman rank correlations with bootstrap confidence intervals and strict false-discovery-rate control across a prespecified family of 16 markers, the researchers found that higher LAR tracked white blood cell count (correlation coefficient 0.406), marrow blast percentage (0.298), and, more weakly, TP53 variant allele frequency (0.217). Among count-derived inflammatory indices, LAR correlated positively with the systemic inflammation response index and the aggregate index of systemic inflammation, and inversely with the platelet-to-lymphocyte and lymphocyte-to-monocyte ratios. But the classic acute-phase proteins told a different story: C-reactive protein, ferritin, and fibrinogen showed no significant correlation with LAR. The pattern, in other words, followed leukocyte and blast burden rather than a uniform systemic inflammatory response.

The outcome associations were striking. Each doubling of pretreatment LAR was associated with a 57 percent higher hazard of death (adjusted hazard ratio 1.57; 95 percent confidence interval 1.20–2.05) and a 44 percent higher hazard of the composite endpoint of progression or death (1.44; 1.14–1.84). These estimates survived adjustment for age, sex, TP53 variant allele frequency, diagnosis, and complex karyotype, and remained essentially unchanged when the researchers additionally accounted for the number of TP53 variants, co-mutation burden, and treatment intensity (fully adjusted OS hazard ratio 1.64; 1.21–2.21). Restricted cubic splines confirmed log-linearity across the observed range, with no evidence of a threshold effect.

But here is where the story takes a turn that biomarker enthusiasts should heed. When the team measured how much LAR actually improved patient-level risk ranking beyond the base clinical model, the optimism-corrected increase in Harrell’s C-index was a mere 0.005 for overall survival and 0.013 for composite progression-free survival. The likelihood-ratio tests showed genuine improvement in model fit, yet the practical gain in discrimination was negligible. Association, the authors emphasize, is not the same as clinically useful prediction—and this distinction matters enormously for anyone hoping to translate a statistical signal into a treatment-selection tool.

The component analyses delivered perhaps the most biologically revealing insight. When lactate dehydrogenase and albumin were entered simultaneously into survival models, LDH per doubling retained a strong association with both overall survival (hazard ratio 1.53) and composite progression-free survival (1.38), whereas albumin per 5-gram-per-liter increment showed no significant association with survival (0.86) and only an imprecise inverse trend for progression-free survival (0.83). The signal, in short, is weighted toward LDH—the enzyme released by rapidly turning-over leukemic cells—rather than toward albumin, the protein that nutrition–inflammation indices typically treat as a proxy for nutritional reserve.

The competing-risk analysis adds yet another layer of technical sophistication. In TP53-mutated disease, early death is common, and death prevents subsequent progression from ever being observed. The researchers therefore separated two fundamentally different questions: the cause-specific rate of progression among patients still alive and progression-free, and the cumulative incidence of progression when death competes. On the cause-specific scale, each LAR doubling was associated with a 39 percent higher progression rate (1.39; 1.05–1.83). But on the Fine–Gray subdistribution scale in the pooled cohort, the association vanished entirely (subdistribution hazard ratio 1.02; 0.80–1.31). The divergence, the authors explain, reflects the event structure rather than model disagreement: patients with higher LAR may progress more rapidly while at risk but also die before progression is documented. Notably, the pooled null estimate was driven by the AML stratum, where competing death is strongest; in MDS, where progression largely means transformation to AML and pre-transformation death is less frequent, LAR was associated with progression on both scales (subdistribution hazard ratio 1.73; 1.14–2.61).

What does all this mean for the nutrition–inflammation framing? The study contains no validated nutritional assessment—no body-mass index, no weight trajectory, no dietary intake, no NRS-2002 or GLIM criteria—so a nutritional contribution to LAR can be neither confirmed nor excluded. Exploratory correlations with cholinesterase (inverse) and triglycerides (positive) hint at possible hepatic synthetic or metabolic contributions, but these analyses were unadjusted for multiplicity and based on small subsets. Similarly, C-reactive protein and ferritin were measured at clinical discretion rather than by protocol, and the measured and unmeasured groups differed systematically in variant allele frequency and albumin, making the missingness informative rather than random. The null acute-phase correlations therefore characterize only what was available, not what might exist.

The authors are refreshingly candid about the interpretive boundaries. LAR, they conclude, should be regarded as a disease-burden-weighted prognostic correlate whose nutritional significance requires direct validation—not as a standalone nutritional index, prediction tool, or treatment-selection marker. The correlations with leukocyte and blast burden, together with the stronger contribution of LDH than albumin, fit a disease-burden interpretation far better than a nutritional one. Experimental work linking mutant p53 to altered glycolysis, marrow lactate accumulation, and histone lactylation provides a tantalizing metabolic context, but circulating total LDH is neither specific to LDHA activity nor a direct measure of lactate production, so the clinical association cannot be assumed to reflect these intracellular processes.

The study’s strengths are methodological rigor: a molecularly defined cohort, continuous modeling with spline checks, false-discovery-rate control, explicit separation of mathematically coupled markers that share albumin with LAR, competing-risk estimands addressed on both cause-specific and subdistribution scales, and bootstrap-corrected discrimination. Its limitations—retrospective design, single-center setting, modest sample size, variant-count surrogate for TP53 allelic state, and lack of external validation—are equally clear. Prospective multicenter studies incorporating copy-number-resolved TP53 classification, LDH isoenzymes, serum and marrow lactate, and direct measures of nutritional status could clarify the source of the signal. Until then, LAR remains a hypothesis-generating number: accessible, reproducible, and undeniably associated with outcome in TP53-mutated myeloid neoplasms, but weighted toward the burden of malignant cells rather than the nutritional state of the patient who carries them.

Subject of Research: A nutrition–inflammation index weighted toward disease burden: the lactate dehydrogenase-to-albumin ratio in TP53-mutated myeloid neoplasms

Article Title: A nutrition–inflammation index weighted toward disease burden: the lactate dehydrogenase-to-albumin ratio in TP53-mutated myeloid neoplasms

Article References: Xin, X., & Wang, Z. (2026). A nutrition–inflammation index weighted toward disease burden: the lactate dehydrogenase-to-albumin ratio in TP53-mutated myeloid neoplasms. Frontiers in Nutrition, 13, Article 1966874. https://doi.org/10.3389/fnut.2026.1966874

Image Credits: AI Generated

DOI: 10.3389/fnut.2026.1966874

Keywords: nutrition, inflammation, index, weighted, toward, disease, burden, lactate, dehydrogenase-to-albumin, ratio, TP53-mutated, myeloid