National Health and Nutrition Examination Survey (NHANES) (16) was a nationally representative cross-sectional survey periodically conducted in the United States by the National Center for Health Statistics, employing a stratified multistage random sampling design. Our retrospective analysis utilized publicly accessible NHANES data spanning from 1999 to 2018.

Within the 1999–2018 NHANES dataset, our scrutiny was confined to 96,811 participants. From this pool, exclusions were made as follows: 41,790 individuals with incomplete cancer information, 49,894 without a cancer diagnosis, 3,693 diagnosed with cancers other than melanoma or non-melanoma skin cancers, 75 with missing follow-up data, 116 lacking essential values such as albumin, BMI, neutrophils, and lymphocytes. Furthermore, an additional 94 were excluded due to missing covariate data. Consequently, the study cohort was consolidated to encompass 1,149 participants ( Figure 1 ).

The ALI was calculated using the formula: BMI (kg/m^2) × serum albumin (g/dL)/NLR. Patients were stratified into quartiles based on their ALI scores, forming four distinct groups: Q1 group (ALI ≤ 37.87), Q2 group (ALI > 37.87 and ≤ 52.84), Q3 group (ALI > 52.84 and ≤ 73.20), and Q4 group (ALI > 73.20).

The primary outcomes of interest were all-cause mortality and cancer mortality. The cause of death was determined using the International Classification of Diseases, 10th Edition (ICD-10) codes. All-cause and cancer mortalities were identified using ICD-10 codes ranging from I00 to I078. For participants included in the NHANES cohort from 1999 to 2018, mortality follow-up data was available up to December 31, 2019.

Demographic variables such as age, gender, and race were self-reported by participants. Laboratory measurements, including alanine aminotransferase (ALT), lactic dehydrogenase (LDH), creatinine (Cr), albumin, neutrophil, and lymphocyte counts, were obtained using standardized automated hematological analysis equipment. The National Center for Health Statistics provided detailed methodologies for these measurements on its website. The BMI was determined by the standard calculation of weight (kg)/[height (m)]^2. Smoking status was categorized based on participant responses; they were identified as never smokers, former smokers, or current smokers according to their smoking history and current smoking habits. Alcohol consumption was assessed and participants were grouped as non-drinkers, former drinkers, mild drinkers, moderate drinkers, or heavy drinkers based on their self-reported alcohol intake. The criteria for hypertension included either taking anti-hypertensive medication or having a mean systolic blood pressure (SBP) ≥ 140 mmHg or a mean diastolic blood pressure (DBP) ≥ 90 mmHg at the time of measurement or a self-reported diagnosis. Diabetes mellitus was defined by the use of hypoglycemic agents or insulin, hemoglobin A1c levels ≥ 6.5%, fasting glucose ≥ 7.0 mmol/L, a random blood glucose ≥ 11.1 mmol/L, a 2-hour oral glucose tolerance test ≥ 11.1 mmol/L, or a self-reported diagnosis. Prediabetes was characterized by fasting blood glucose levels between 6.0-7.0 mmol/L or 2-hour postprandial glucose levels between 7.8-11.1 mmol/L. A history of cardiovascular disease (CVD) was determined by self-reported history of conditions such as congestive heart failure, heart attack, coronary heart disease, angina, or stroke.

We used the NHANES-recommended weights to calculate the appropriate weights for specific groups. Continuous variables were presented as mean ± standard deviation, and for those not following a normal distribution, we represented them by the median (25th percentile, 75th percentile). Categorical variables were reported as counts (percentages). To compared baseline characteristics among the four groups, we applied variance analysis (ANOVA) for continuous variables and the χ2 test for categorical variables.

In analyzing the association between ALI and skin cancer all-cause mortality and cancer mortality, our analysis included Kaplan-Meier and Cox regression analyses, utilizing the NHANES-recommended weights. Model 1 adjusted for demographic factors: age (years), gender (male or female), and race (White, Black, Mexican American, or other). Enhancing Model 1, Model 2 incorporated adjustmented for smoke status (never, former, or current), alcohol consumption (never, former, mild, moderate, or heavy), and disease status (presence or absence, including history of hypertension, diabetes mellitus (DM), and CVD). Further refining our analysis, Model 3 added ALT (U/L), LDH (mmol/L), and Cr (umol/L) to the adjustments made in Model 2. To identify potential non-linear relationships between ALI and all-cause mortality and cancer mortality, we employed restricted cubic splines (RCS). An inflection point was determined from the RCS analysis, and its impact was assessed using segmented Cox analysis. To quantify the ALI’s impact, we divided the ALI levels of each participant by 10, assessing the effect of every 10-unit change in ALI on all-cause mortality and cancer mortality in skin cancer patients. Subsequent COX regression analysis was performed on the variables required for ALI calculation. We also undertooked a stratified analysis to explore the ALI and skin cancer all-cause and cancer mortality relationship across different subgroups, including age, gender, smoke status, hypertension, DM, ALT, and Cr, enhancing the robustness of our findings.

All analyses were conducted using R software (version 4.3.1), with a two-sided P-value of <0.05 designated as the threshold for statistical significance in all analyses.

Among all 1149 participants who met the study criteria, the average age was 63.13 (62.14, 64.13). The proportion of males was higher (n = 653, 56.83%), with the majority being White (n = 1075, 93.56%). Based on ALI quartiles, patients were divided into four groups: Q1 (n = 288), Q2 (n = 286), Q3 (n = 287), and Q4 (n = 288). The ALI median values for Q4 (105.64), Q3 (45.11), and Q2 (61.61) were higher than for Q1 (29.39).

Participants in the higher ALI groups were younger (Q1: 68.04 vs. Q2: 63.80 vs. Q3: 61.89 vs. Q4: 59.93) and had higher BMI (Q1: 25.64 vs. Q2: 27.13 vs. Q3: 28.78 vs. Q4: 30.88). In the higher ALI groups, participants had lower neutrophil levels (Q1: 5.34 vs. Q2: 4.49 vs. Q3: 4.00 vs. Q4: 3.40) but higher lymphocyte levels (Q1: 1.44 vs. Q2: 1.77 vs. Q3: 1.99 vs. Q4: 2.91). Statistical differences in albumin, ALT, and Cr were also observed across the groups. However, other indicators, including the number of gender, race, LDH, smoking status, alcohol consumption, disease status, and the proportion of skin cancer among the groups, showed no statistically significant difference. More data on the baseline characteristics of the study population can be found in Table 1 .

Among the 1,149 skin cancer patients included, there were 234 (20.37%) with melanoma, 615 (53.52%) with non-melanoma, and 300 (26.11%) with Skin (unknown). These were divided into Q1, Q2, Q3, and Q4 groups using quartile methods for Kaplan-Meier survival analysis curves. From the graph, we could see that ALI was correlated with both all-cause mortality and cancer mortality in skin cancer patients (all P-values <0.05, Figure 2 ).

The univariate Cox proportional hazard results ( Table 2 ) showed that, compared to the Q1 group, the all-cause mortality risk in skin cancer patients in the Q2 group (HR: 0.54, 95% CI: 0.38–0.77), Q3 group (HR: 0.30, 95% CI: 0.21–0.43), and Q4 group (HR: 0.31, 95% CI: 0.22–0.44) decreased by 46%, 70%, and 69%, respectively. This difference was statistically significant (P for trend <0.001).

After adjusting for potential confounding factors such as age, gender, race, smoking status, alcohol consumption, hypertension, DM, CVD, ALT, LDH, and Cr, compared to the Q1 group, the all-cause mortality risk in skin cancer patients in the Q2 group (HR: 0.77, 95% CI: 0.57–1.05), Q3 group (HR: 0.50, 95% CI: 0.36–0.69), and Q4 group (HR: 0.54, 95% CI: 0.38–0.77) still showed varying degrees of reduction. This difference was statistically significant (P for trend <0.001).

Similarly, both the unadjusted and adjusted Cox proportional hazard results showed that, compared to the Q1 group, the cancer mortality risk in skin cancer patients in the Q2, Q3, and Q4 groups all decreased to varying degrees. This difference was statistically significant (P for trend = 0.005 and P for trend = 0.049).

Through restricted cubic splines analysis combined with the Cox proportional hazards model, we found a reverse J-shaped non-linear relationship between ALI and all-cause mortality and cancer mortality in skin cancer, as shown in Figure 3 . Our results indicated that the inflection point for all-cause mortality in skin cancer was 81.13, and for cancer mortality, it was 77.50.

When the ALI was above the inflection point, an increase of 10U in ALI resulted in a 2% and 6% increased in the multivariate-adjusted HR for all-cause mortality and cancer mortality, respectively (HR 1.02, 95% CI: from 1.00 to 1.04 and HR 1.06, 95% CI: from 1.05 to 1.07). On the other hand, when the ALI was below the inflection point, an increase of 10U in ALI resulted in a 20% and 21% decrease in the multivariate-adjusted HR for all-cause mortality and cancer mortality, respectively (HR 0.80, 95% CI: from 0.74 to 0.86 and HR 0.79, 95% CI: from 0.66 to 0.95), as shown in Table 3 .

When participants were stratified by age (P for interaction = 0.12), gender (P for interaction = 0.22), smoking status (P for interaction = 0.63), hypertension (P for interaction = 0.43), DM (P for interaction = 0.67), ALT (P for interaction = 0.67), and Cr (P for interaction = 0.15), the association between ALI and all-cause mortality did not change. Sensitivity analysis showed that among individuals with Cr levels ≥106 umol/L, those with ALI greater than 81.37 had a 68% reduced risk of all-cause mortality in skin cancer patients (HR 0.32, 95% CI: from 0.14 to 0.87, P for trend = 0.01). No other indicators had significant associations. Similarly, when participants were stratified by age (P for interaction = 0.27), gender (P for interaction = 0.71), smoking status (P for interaction = 0.39), hypertension (P for interaction = 0.65), DM (P for interaction = 0.63), and Cr (P for interaction = 0.59), the association between ALI and cancer mortality did not change. Sensitivity analysis showed that the results for ALI and cancer mortality in skin cancer were consistent with the main effects (Additional file 1).