Study participants and baseline characteristics
The baseline characteristics are summarized in Table 1. The average age was 47.46 years (50.29% male, 49.71% female). The cancer patients had a median age of 62.46 years, significantly older than the non-cancer participants (45.81 years). Higher cancer incidence was noted among non-Hispanic whites, those with education level of high school or above, married or cohabiting individuals, alcohol consumers, PIR > 1, and those with hypertension, hyperlipidemia, low albumin levels, and higher NPAR. The average BMI was 28.95 kg/m2. Notably, albumin levels were higher and neutrophil proportion and NPAR lower in the non-cancer group.
Table 1
Baseline characteristics of study participants in NHANES 2005–2016
Variables | Total participant, N = 25,026 | Non-cancer, N = 22,650 | Cancer, N = 2,376 | p value |
Age, years | 47.46 (0.26) | 45.81 (0.26) | 62.46 (0.39) | < 0.0001 |
Gender | | | | < 0.0001 |
Male | 12586 (50.29) | 11457 (50.19) | 1129 (43.44) | |
Female | 12440 (49.71) | 11193 (49.81) | 1247 (56.56) | |
Ethnicity | | | | < 0.0001 |
Non-Hispanic white | 11521 (46.04) | 9824 (68.56) | 1697 (87.82) | |
Non-Hispanic black | 5041 (20.14) | 4728 (10.76) | 313 (4.69) | |
Mexican American | 3888 (15.54) | 3744 (8.57) | 144 (2.12) | |
Other race | 4576 (18.28) | 4354 (12.11) | 222 (5.37) | |
Education | | | | < 0.0001 |
Below high school level | 6074 (24.27) | 5584 (16.21) | 490 (12.38) | |
High school | 5743 (22.95) | 5213 (22.77) | 530 (20.70) | |
Above high school | 13209 (52.78) | 11853 (61.02) | 1356 (66.92) | |
Marital | | | | < 0.0001 |
Married/living with partner | 15027 (60.05) | 13581 (64.03) | 1446 (65.68) | |
Widowed/divorced/separated | 5591 (22.34) | 4805 (17.45) | 786 (28.52) | |
Never married | 4408 (17.61) | 4264 (18.52) | 144 (5.80) | |
Drinking | | | | < 0.0001 |
Never | 3492 (13.95) | 3186 (10.93) | 306 (10.21) | |
Former | 4600 (18.38) | 4014 (14.73) | 586 (19.92) | |
Current | 16934 (67.67) | 15450 (74.34) | 1484 (69.86) | |
Smoking | | | | < 0.0001 |
Never | 13498 (53.94) | 12441 (55.02) | 1057 (45.42) | |
Former | 6240 (24.93) | 5296 (23.74) | 944 (38.46) | |
Current | 5288 (21.13) | 4913 (21.24) | 375 (16.12) | |
BMI (kg/m2) | 28.95 (0.08) | 28.96 (0.09) | 28.87 (0.15) | 0.57 |
Poverty-to-income ratio | | | | < 0.0001 |
Poor (≤ 1) | 524 5(20.96) | 4897 (14.36) | 348 (9.25) | |
Not poor (> 1) | 19781 (79.04) | 17753 (85.64) | 2028 (90.75) | |
Hypertension | | | | < 0.0001 |
No | 14387 (57.49) | 13531 (64.44) | 856 (42.44) | |
Yes | 10639 (42.51) | 9119 (35.56) | 1520 (57.56) | |
Hyperlipidemia | | | | < 0.0001 |
No | 7007 (28) | 6601 (30.01) | 406 (16.97) | |
Yes | 18019 (72) | 16049 (69.99) | 1970 (83.03) | |
Diabetes | | | | < 0.0001 |
No | 20392 (81.48) | 18662 (87.01) | 1730 (78.09) | |
Yes | 4634 (18.52) | 3988 (12.99) | 646 (21.91) | |
NPAR | 13.69 (0.03) | 13.64 (0.03) | 14.21 (0.07) | < 0.0001 |
Q1 | 6250 (24.97) | 5821 (24.73) | 429 (18.46) | |
Variables | Total participant, N = 25,026 | Non-cancer, N = 22,650 | Cancer, N = 2,376 | p value |
Q2 | 6256 (25) | 5759 (26.65) | 497 (22.85) | |
Q3 | 6262 (25.02) | 5632 (25.38) | 630 (27.78) | |
Q4 | 6258 (25.01) | 5438 (23.24) | 820 (30.91) | |
Albumin, g/dL | 4.30 (0.00) | 4.30 (0.00) | 4.23 (0.01) | < 0.0001 |
Neutrophil percent, % | 58.44 (0.11) | 58.30 (0.11) | 59.71 (0.25) | < 0.0001 |
Neutrophil count, 109/L | 4.30 (0.02) | 4.30 (0.02) | 4.32 (0.05) | 0.65 |
Lymphocyte count, 109/L | 2.13 (0.01) | 2.14 (0.01) | 2.07 (0.03) | 0.04 |
Note: The continuous variables were analyzed by Wilcoxon test, expressed by the median (IQR); the weighted chi-square test was used to analyze the categorical variables, expressed by the column percentage. A level of two-sided p < 0.05 was considered statistically significant. Means and standard error were described for the continuous variables, counts and proportions (after weighted) were described for categorical variables. Abbreviations: BMI, body mass index; IQR, Interquartile range; NHANES, National Health and Nutrition Examination Survey; NPAR, neutrophil-percentage-to-albumin ratio. |
Association of NPAR with cancer risk: NHANES 2005–2016
The association between NPAR and cancer risk was analyzed using weighted multifactor Logistic regression, both as a categorical and continuous variable. The findings are detailed in Table S1. Initial univariate logistic regression revealed an increased likelihood of cancer in higher NPAR quartiles: the odds increased by 47% (OR = 1.47 [95% CI 1.20, 1.79]) for the third quartile (Q3) and 78% (OR = 1.78 [95% CI 1.49, 2.13]) for the fourth quartile (Q4), compared to the first quartile (Q1). A continuous variable analysis of NPAR also demonstrated a positive a positive correlation with cancer risk (OR = 1.10 [95% CI 1.07, 1.12]). Subsequent multivariable logistic regression, adjusted for factors like age, sex, race, smoking, alcohol consumption, poverty-to-income ratio, hypertension, hyperlipidemia, and diabetes (Table 2), yielded odds ratios for Q2, Q3, and Q4 of 0.97 (95% CI 0.76, 1.25), 0.99 (95% CI 0.69, 1.41), and 0.98 (95% CI 0.60, 1.59), respectively. Continuous analysis under this model showed similar findings (OR = 1.07 [95% CI 0.84, 1.36]). Notably, p-values in both categorical and continuous analyses were above 0.05, suggesting no statistically significant association.
Table 2
The relationship between NPAR and mortality among cancer patients in NHANES 2005–2016
Cancer | Character | Range median (IQR) | aHR a (95% CI) | p value |
Pan-cancer | NPAR | | 1.08(1.04,1.12) | < 0.0001 |
| Q1 | 11.43(0.86–12.67) | ref | ref |
| Q2 | 13.55(12.67–14.31) | 1.22(0.92,1.62) | 0.16 |
| Q3 | 15.12(14.31–16.05) | 1.16(0.89,1.51) | 0.28 |
| Q4 | 17.41(16.05–36.10) | 1.73(1.34,2.22) | < 0.0001 |
| P for trend | | < 0.0001 |
Thoracic cancer | NPAR | | 1.07(0.99,1.16) | 0.08 |
| Q1 | 11.468(6.95–12.93) | ref | ref |
| Q2 | 13.884(12.93–14.70) | 1.18(0.60,2.29) | 0.63 |
| Q3 | 15.366(14.70-16.56) | 1.07(0.59,1.96) | 0.82 |
| Q4 | 18.244(16.56–29.47) | 1.50(0.80,2.82) | 0.21 |
| P for trend | | 0.232 |
Abdominal cancer | NPAR | | 1.14(1.04,1.26) | 0.004 |
| Q1 | 11.67(4.16–12.91) | ref | ref |
| Q2 | 13.89(12.91–14.39) | 1.09(0.37,3.18) | 0.88 |
| Q3 | 15.05(14.39–16.11) | 1.54(0.74,3.21) | 0.25 |
| Q4 | 17.60(16.11–28.07) | 2.61(1.30,5.24) | 0.01 |
| P for trend | | 0.005 |
Reproductive system cancer | NPAR | | 1.07(0.90,1.27) | 0.46 |
| Q1 | 11.45(8.28–12.67) | ref | ref |
| Q2 | 13.52(12.67–13.98) | 1.10(0.30, 4.04) | 0.88 |
| Q3 | 14.84(13.98–15.73) | 0.62(0.11, 3.57) | 0.6 |
| Q4 | 16.77(15.73–21.79) | 0.89(0.24, 3.32) | 0.87 |
| P for trend | | 0.718 |
Skin cancer | NPAR | | 1.14(1.05,1.24) | 0.002 |
| Q1 | 11.55(2.62–12.51) | ref | ref |
| Q2 | 13.39(12.51–14.16) | 1.25(0.71,2.23) | 0.44 |
| Q3 | 15.05(14.16–15.91) | 1.52(0.91,2.53) | 0.11 |
| Q4 | 16.93(15.91–36.10) | 1.70(1.04,2.77) | 0.03 |
| P for trend | | 0.02 |
Urinary system cancer | NPAR | | 1.07(0.97,1.17) | 0.18 |
| Q1 | 11.52(1.35–12.75) | ref | ref |
| Q2 | 13.68(12.75–14.46) | 0.90(0.53,1.55) | 0.71 |
| Q3 | 15.13(14.46–16.24) | 1.10(0.64,1.88) | 0.73 |
| Q4 | 18.07(16.24–22.35) | 1.53(0.88,2.63) | 0.13 |
| P for trend | | 0.048 |
Note: Cox proportional hazards regression for the relationship between NPAR and mortality among cancer patients. aHR a: Adjusted for age, gender, ethnicity, education level, marital, drinking status, smoking status, Poverty-to-income ratio, hypertension, hyperlipidemia and diabetes. Abbreviations: HR, hazard ratio; CI, confidence interval; IQR, Interquartile range; NPAR, neutrophil percentage-to-albumin ratio. |
Association of NPAR with cancer mortality: NHANES 2005–2016
Follow-up duration was calculated in person-months from the interview date to either the date of death or end of the follow-up period. During the 11-year follow-up, 2376 cancer-related deaths were recorded. Basic demographic and clinical data of cancer patients stratified by NPAR quartiles are detailed in Table S2. Cancer types were classified into categories: thoracic (breast, lung, esophageal), abdominal (colorectal, stomach, liver, gallbladder, pancreatic), reproductive system (uterine, cervical, ovarian, testicular), urinary system (kidney, prostate, bladder), and skin (melanoma and other skin tumors). Adjusted analyses revealed a positive correlation between NPAR and cancer mortality. The relationship was examined in both continuous and categorical formats, assessing mortality rates across various cancer types: overall, thoracic, abdominal, reproductive system, skin, and urinary system tumors (detailed in Table 2 and Fig. 2). In the analysis of overall cancer mortality, continuous NPAR showed an adjusted hazard ratio (aHR) of 1.08 (95% CI 1.04, 1.12); while categorical analysis indicated an increased risk in Q4 compared to Q1 with an aHR of 1.73 [95% CI 1.34, 2.22]), both significant (p < 0.05). Thoracic tumor analysis revealed an aHR of 1.07 (95% CI 0.99, 1.16) for continuous NPAR, and no significant differences in categorical among Q2 (aHR 1.18, 95% CI 0.60, 2.29), Q3 (aHR 1.07, 95% CI 0.59, 1.96), and Q4 (aHR 1.50, 95% CI 0.80, 2.82) against Q1. For abdominal tumors, continuous NPAR analysis showed an aHR of 1.14 (95% CI 1.04, 1.26). Categorically, Q4 showed a significantly higher risk than Q1 (aHR = 2.61 [95% CI 1.30, 5.24]), both p < 0.05. In reproductive system tumors, aHR were 1.07 (95% CI 0.90, 1.27) and 0.89 (95% CI 0.24, 3.32), with no significant differences. Skin tumor analysis showed a 14% increased risk for continuous NPAR (aHR = 1.14 [95% CI 1.05, 1.24]) and a 70% increase for Q4 compared to Q1 in categorical analysis (aHR = 1.70 [95% CI 1.04, 2.77]), both statistically significant. Urinary system tumors analysis found an aHR of 1.07 (95% CI 0.97, 1.17) for continuous NPAR and an increased, but not significant, risk for Q4 compared to Q1 (aHR = 1.53 [95% CI 0.88, 2.63]). Trend tests were also conducted for various tumor types. A significant upward trend in cancer mortality risk was observed with increasing NPAR levels in overall cancer, abdominal, skin, and urinary system tumors, with respective p-values for trend as follows: <0.0001, < 0.005, < 0.02, < 0.048 (Table 2).
Beyond linear analysis, the study also investigated potential non-linear relationship between NPAR and cancer mortality (Fig. 3). Utilizing RCS analysis for model adjustment revealed a significant non-linear relationship between NPAR and overall cancer mortality (p < 0.0001). Specifically, cancer mortality risk decreased to its lowest at an NPAR level of 13.48779, then increased as NPAR rose beyond this point (p < 0.0001). This non-linear trend was also evident in skin cancer and urinary system tumors (p < 0.05). For NPAR levels below 13.38739 and 13.89568, cancer mortality risk declined with rising NPAR; however, exceeding these thresholds led to increased mortality risk. This non-linear pattern was not observed in the other tumor types. Figure 4 displays significant survival probability differences across NPAR quartiles (Q1, Q2, Q3, Q4) using Kaplan-Meier survival analysis.
To corroborate these findings, a correlation study was performed using NPAR data from clinical patients in our hospital. Data were gathered from 49 patients with infiltrating moderately and poorly differentiated gastric adenocarcinoma (2019 to 2020). These patients were categorized into high and low NPAR groups based on the median NAPR value (median = 15.72815534). The relationship between NPAR levels and patient survival prognosis was analyzed and is illustrated in Fig. 5. Patients with higher NPAR exhibited significantly poorer prognosis compared to those with lower NPAR (p = 0.0184).