Is my retrospective cross‑sectional protocol assessing neutrophil‑to‑lymphocyte ratio to predict malnutrition risk (Nutritional Risk Screening 2002 ≥ 3) in adult cancer inpatients at a tertiary government hospital appropriate, and how can it be improved?

Protocol Assessment: NLR as a Malnutrition Predictor in Cancer Patients

Your protocol is methodologically sound but scientifically misaligned with current evidence—NLR measures systemic inflammation, not malnutrition, and NRS-2002 already incorporates disease severity (which reflects inflammation), creating circular validation where you are testing an inflammation marker against a tool that already accounts for inflammatory burden. 1

Critical Conceptual Flaws

NLR Reflects Inflammation, Not Nutritional Status

• Neutrophil-to-lymphocyte ratio is an established marker of systemic inflammatory response in cancer, driven by tumor-released proinflammatory cytokines that stimulate neutrophil production and suppress lymphocyte counts—it does not directly measure nutritional depletion. 2
• Cancer patients exhibit upregulated innate immune responses involving neutrophils and macrophages, which elevate NLR independent of caloric or protein intake. 2
• The 2025 Advances in Nutrition guideline panel explicitly states that NRS-2002 is more specific than inflammatory markers for identifying malnutrition in cancer patients, making NRS-2002 the appropriate first-line tool rather than NLR. 1

Circular Validation Problem

• NRS-2002 already incorporates disease severity as one of its four core components (alongside BMI, weight loss, and reduced intake), and disease severity directly reflects the inflammatory burden that drives NLR elevation. 3, 1
• Testing whether NLR predicts NRS-2002 scores creates a tautology: you are correlating an inflammation marker with a screening tool that already quantifies inflammation through its disease-severity component. 1
• ESPEN guidelines validated NRS-2002 by retrospectively analyzing 128 randomized controlled trials, demonstrating that patients meeting NRS-2002 risk criteria had higher likelihood of positive clinical outcomes from nutritional support—this validation was based on clinical endpoints (complications, mortality, length of stay), not inflammatory biomarkers. 3

What the Existing Research Actually Shows

Studies Demonstrate Association, Not Causation

• A 2021 cross-sectional study of 119 unselected cancer patients found NLR ≥5.0 was associated with higher NRS-2002 scores (3.0±1.1 vs. 2.3±1.2, p=0.0004) and higher prevalence of nutritional risk (73.6% vs. 37.9%, p=0.001), but this association reflects shared inflammatory pathophysiology rather than NLR's ability to independently detect malnutrition. 4
• A 2019 geriatric outpatient study (n=95) found NLR >1.81 predicted malnutrition/risk with 71.7% sensitivity and 63.3% specificity, but the optimal cut-point of 1.81 is far lower than typical cancer-related NLR elevations, limiting applicability to your population. 5
• A 2024 study of 1234 ICI-treated cancer patients demonstrated that NLR's predictive value for treatment outcomes disappeared entirely when serum albumin was <3.8 g/dL, proving that malnutrition confounds NLR interpretation rather than NLR detecting malnutrition. 6

Geriatric Data Cannot Be Extrapolated to Cancer Populations

• A 2023 geriatric inpatient study (n=220, mean age 77.9 years) found NLR >4.5 predicted malnutrition with 37.9% sensitivity and 85.2% specificity, but the optimal cut-point and predictive performance differ substantially from cancer cohorts due to distinct inflammatory profiles. 7
• In typical hospitalized cancer cohorts, malnutrition prevalence ranges from 40-50%; even modest specificity generates substantial false-positive results, limiting clinical utility. 1

How to Improve Your Protocol

Reframe the Research Question

• Instead of asking whether NLR predicts NRS-2002 scores, investigate whether NLR adds incremental prognostic value beyond NRS-2002 for predicting clinical outcomes (postoperative complications, 30-day mortality, length of stay, readmission rates). 1
• This reframing aligns with ESPEN's validation methodology, which linked nutritional screening to hard clinical endpoints rather than cross-validating screening tools against each other. 3

Strengthen the Study Design

• Convert from cross-sectional to prospective cohort design with standardized NRS-2002 assessment at admission, concurrent NLR measurement, and follow-up for clinical outcomes—this captures the dynamic changes in both inflammation and nutritional status during active cancer treatment. 1
• Include longitudinal NLR measurements at days 3,7, and 14 to assess whether NLR trajectory correlates with nutritional deterioration or recovery. 1

Expand Outcome Measures Beyond NRS-2002

• Primary outcomes should include postoperative complications (Clavien-Dindo grade ≥3), 30-day mortality, hospital length of stay, and 90-day readmission rates—these are the clinically meaningful endpoints that justify nutritional screening. 3, 1
• Secondary outcomes can include change in NRS-2002 score from admission to discharge, actual nutritional intervention rates, and compliance with prescribed nutritional support. 1

Refine Exclusion Criteria

• Your current exclusion of "acute or chronic infections" and "autoimmune diseases" will eliminate a substantial proportion of cancer patients, reducing generalizability. 4
• Consider stratifying by infection/inflammation status rather than excluding these patients, allowing subgroup analysis of NLR performance in high- vs. low-inflammation states. 6
• Exclude only patients receiving granulocyte colony-stimulating factors within 7 days of CBC collection, as these directly manipulate neutrophil counts. 4

Add Complementary Inflammatory Markers

• Include C-reactive protein (CRP) and albumin to calculate the modified Glasgow Prognostic Score (mGPS), which is highly predictive of morbidity and mortality in cancer patients and provides context for NLR interpretation. 3, 2
• Median CRP values in solid tumors are approximately 46 mg/L, substantially higher than cardiovascular disease (6 mg/L) but lower than bacterial infections (120 mg/L), allowing differentiation of cancer-related inflammation from acute infection. 2

Methodological Strengths to Preserve

Appropriate Screening Tool Selection

• NRS-2002 is the ESPEN-recommended validated screening tool for hospitalized patients, including surgical and critically ill cancer patients, with documented predictive validity for clinical outcomes. 3, 1
• NRS-2002 can be performed without laboratory tests, requiring only BMI, weight history, food-intake assessment, and disease-severity information, making it feasible in resource-limited settings. 1

Adequate Sample Size Considerations

• Your retrospective chart review design is appropriate for hypothesis generation, but power calculations should target the rarest outcome of interest (likely 30-day mortality at approximately 5-10% in mixed cancer populations). 1
• Plan for at least 200-300 patients to allow multivariable logistic regression with 10-15 events per predictor variable. 4, 7

Robust Statistical Approach

• Binary logistic regression and ROC curve analysis are appropriate for assessing NLR's discriminative ability, but ensure you adjust for known confounders (cancer type, stage, treatment status, comorbidities, CRP, albumin). 4, 5
• Report both sensitivity/specificity and positive/negative predictive values, as the latter depend on malnutrition prevalence in your specific population. 7

What Current Guidelines Actually Recommend

Nutritional Screening in Cancer Patients

• ESPEN expert group recommendations (2017) advise screening for nutritional risk as soon as cancer is diagnosed, followed by full nutritional assessment when risk is present, using validated tools like NRS-2002 rather than inflammatory biomarkers. 3
• All patients in contact with health or elderly care should undergo nutritional risk screening as the first mandatory step, with re-screening every 7-10 days during hospitalization. 3, 1

Role of Inflammatory Markers

• The Glasgow Prognostic Score (GPS), based on CRP and albumin, is an easy-to-use and highly predictive tool for assessing inflammation and prognosis in cancer patients, but it is used for prognostic stratification, not malnutrition screening. 3
• Poor cancer outcomes are predicted by markers of systemic inflammatory response (elevated CRP, hypoalbuminemia, elevated white cell counts), but these reflect disease burden rather than nutritional status per se. 2

Common Pitfalls to Avoid

• Do not present NLR as a "cost-effective alternative" to NRS-2002—NRS-2002 requires no laboratory tests and is therefore already more cost-effective than any blood-based marker. 1
• Do not claim NLR "detects malnutrition" when it actually detects inflammation; the correct framing is that NLR may identify cancer patients at high risk for poor outcomes who warrant intensive nutritional assessment. 2, 6
• Do not ignore the 2024 evidence showing that malnutrition (albumin <3.8 g/dL) abolishes NLR's predictive value—this suggests malnutrition is a confounder of NLR interpretation rather than an outcome NLR can detect. 6
• Avoid using NLR cut-points derived from geriatric populations (1.81-4.5) for cancer patients, as cancer-related inflammation typically produces higher NLR values (≥5.0). 4, 7, 5