Abstract
Background
Cancer causes a serious health burden on patients worldwide. Chronic low‐level inflammation plays a key role in tumorigenesis and prognosis. However, the role of the red blood cell distribution width (RDW)‐to‐albumin (RA) ratio in cancer mortality remains unclear.
Methods
In this retrospective cohort study, we collected clinical information from cancer patients from the Medical Information Mart for Intensive Care III (MIMIC‐III) version 1.4 database and then calculated RA by dividing RDW by albumin concentration. The primary outcome was 30 days mortality, while secondary outcomes were 90 days and 1 year mortality. Next, we adopted Cox regression models to calculate hazard ratios (HR) together with 95% confidence intervals (CI) for all‐cause mortalities associated with the RA ratio.
Results
For 30 days mortality, the HR (95% CI) for the high RA ratio (≥5.51) was 2.17 [95CI% (1.87–2.51); p = <0.0001], compared with the low RA ratio (<5.51). In Model 2, we adjusted sex and age and obtained HR (95% CI) of 2.17 [95CI% (1.87–2.52); p = <0.0001] for the high RA ratio (≥5.51) group, compared to that in the low RA ratio (<5.51). In Model 3, adjusting for age, sex, anion gap, hematocrit, white blood cell count, congestive heart failure, SOFA, liver disease, and renal failure resulted in HR (95% CI) of 1.74 [95CI% (1.48–2.04); p = <0.0001] for the high RA ratio (≥5.51) relative to the low RA ratio (<5.51). We also analyzed common diseases in cancer patients but found no significant association.
Conclusion
To the best of our knowledge, this is the first study demonstrating that increased RA ratio is independently associated with increased all‐cause mortality in cancer patients.
Keywords: albumin, cancer, medical information mart for intensive care‐III, RDW
Cancer causes a serious health burden on patients worldwide. Chronic low‐level inflammation plays a key role in tumorigenesis and prognosis. we provide the first evidence showing that increased RA is an independent predictor of increased all‐cause mortality in cancer patients.
1. INTRODUCTION
Cancer causes a serious health burden on patients worldwide. 1 , 2 Previous estimates have shown that the global incidence of cancer will increase from 12.7 to 22.2 million cases by 2030. 3 Overall cancer mortality has declined owing to advancement in techniques used for the early detection of tumors and the emergence of new treatment strategies. 4 , 5 , 6 However, the number of patients admitted to the intensive care unit (ICU) has increased. 7 Results from a previous epidemiological study revealed an ICU admission rate of 5.2% within 2 years of a definite diagnosis of cancer. 8 Currently, identification of biomarkers for predicting the prognosis of cancer patients in the ICU is a hot research topic. 9
Previous studies have shown that chronic low‐level inflammation plays a significant role in both tumorigenesis and prognosis. 10 , 11 The red blood cell distribution width (RDW), which can be obtained by evaluating complete blood count, is used to represent variability in the size of circulating erythrocytes and differentiate different types of anemia in clinical settings. Recent studies have demonstrated the ability of RDW to reflect inflammation and nutritional dysregulation in lung cancer patients with cardiovascular disease, 12 , 13 , 14 sepsis, 15 and chronic obstructive pulmonary disease (COPD). 16 Additional studies have shown that RDW is a new and effective indicator of the general condition and mortality of patients with lung cancer. In patients with breast cancer, elevated RDW exhibited a significant positive correlation with the size of primary tumors, degree of axillary lymphatic spread, and levels of HER2 expression, but was negatively correlated with tumor grade. 17 Some scholars believe that RDW can be used as a novel indicator of tumor metastasis in breast and other solid tumors, owing to its advantages of convenience and cost‐effectiveness. 18 , 19 To date, however, RDW’s prognostic value in patients with tumors admitted to the ICU remains unknown. A recent study proposed analyzed the use of RDW in combination with other identified biomarkers, such as serum albumin level, and found that it was associated with mortality of patients with various diseases including tumors. 20 Albumin not only exerts anti‐inflammatory effects but also reduces oxidative stress and inhibits apoptosis of endothelial cells. 21 , 22 The RDW‐to‐albumin (RA) ratio is a novel inflammatory biomarker, 23 which has previously been used to assess the prognosis of patients with stroke 22 and aortic aneurysms. 23 , 24 We hypothesized that the use of RDW in combination with albumin may be a potential predictor of cancer mortality. In the present study, we assessed the prognostic value RA ratio in predicting cancer mortality.
2. MATERIALS AND METHODS
2.1. Study population
Clinical data for 50,000 critically ill patients, who were admitted to the Beth Israel Deaconess Medical Center between 2001 and 2012, 25 were obtained from a free accessible critical care Medical Information Mart for Intensive Care III database version 1.4 (MIMIC‐III, v1.4). To access the database, we completed the Protecting Human Research Participants, an online course developed by the National Institutes of Health. The database was recognized by the institutional review boards of the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center. All personal information was removed to protect the privacy of patients. Patients were included in the study if they: (1) were diagnosed with cancer based on the International Classification of Diseases, Ninth Revision (ICD‐9); (2) were aged ≥16 years; and (3) only had one ICU admission during the study period. Conversely, subjects who had more than 10% of the data missing and those with a length of hospital stay <24 h were excluded from the study.
2.2. Study variables
Study variables included demographic characteristics (age, gender, race), vital signs, laboratory indices, and comorbidities. Vital signs included heart rate, oxygen saturation (SPO2), systolic blood pressure (SBP), diastolic blood pressure (DBP), mean arterial pressure (MAP), respiratory rate, and body temperature. Comorbidities included acute coronary artery disease (CAD), kidney disease, and liver disease, whereas laboratory indices included neutrophil count, monocyte count, lymphocyte count, white blood cell (WBC) count, hemoglobin, platelet count, RDW, glucose level, serum creatinine level, blood urea nitrogen (BUN), and anion gap within 24 h of ICU admission. Sequential Organ Failure Assessment (SOFA) was also included. Primary outcome was all‐cause 30 days mortality, whereas secondary outcomes were all‐cause 90 days and 1 year mortality, and the length of ICU stay.
2.3. Statistical analysis
Continuous variables that conformed to normal distribution were presented as means ± standard deviations (SD), non‐normally distributed data were presented using medians (interquartile range). Between‐group differences were determined using the Wilcoxon W or Kruskal‐Wallis tests. Categorical variables were expressed as numbers and percentages and then compared between groups using the chi‐squared or Fisher's exact tests. The RA ratio was evaluated in the tertile and dichotomized groups, with that in the former group considered the reference value. Multivariate analysis was performed using Cox regression models and used to investigate the prognostic value of the RA ratio in predicting cancer mortality. Confounding factors, with estimated effects >10%, were selected for adjustment. 26 These included age, sex, anion gap, hematocrit, white blood cell count, congestive heart failure, SOFA scores, liver disease, and renal failure. In addition, we performed a stratified analysis based on each variable and comorbidity to examine the stability of RA in predicting disease survival outcomes across various subgroups. Moreover, propensity score matching was performed because of the differences in baseline characteristics. Propensity score matching, at a ratio of 1:1, was performed using a caliper width of 0.01 of the SD of the logit of the propensity score. 27 All statistical analyses were performed using packages implemented in R software version 4.01 (https://www.r‐project.org/), and statistical significance was defined by a two‐tailed p‐value <0.05.
3. RESULTS
3.1. Subject characteristics
A total of 3381 cancer patients were enrolled in this study. Details on their demographic characteristics, vital signs, laboratory indices, and comorbidities at baseline are outlined in Table 1. Summarily, patients with higher RA ratios had higher 30 days, 90 days, and one year mortality, but lower SBP, DBP, and MAP. In addition, this patient population had a history of renal failure and liver disease.
TABLE 1.
Baseline characteristics of the study population
| Characteristics | Total | RA Ratio | p value | |
|---|---|---|---|---|
| <5.51 | ≥5.51 | |||
| N | 3381 | 1686 | 1695 | |
| Age, years | 65.78 ± 14.44 | 65.85 ± 14.72 | 65.73 ± 14.18 | 0.803 |
| Sex, n (%) | 0.407 | |||
| Female | 1356 (40.11) | 688 (40.81) | 668 (39.41) | |
| Male | 2025 (59.89) | 998 (59.19) | 1027 (60.59) | |
| Vital signs | ||||
| SBP, mmHg | 116.48 ± 17.19 | 120.11 ± 17.07 | 112.87 ± 16.54 | <0.001 |
| DBP, mmHg | 60.82 ± 10.90 | 62.50 ± 10.83 | 59.16 ± 10.72 | <0.001 |
| MAP, mmHg | 77.01 ± 11.47 | 78.98 ± 11.30 | 75.05 ± 11.31 | <0.001 |
| Heart rate, beats/min | 89.89 ± 17.17 | 87.09 ± 16.89 | 92.69 ± 17.00 | <0.001 |
| SpO2, % | 96.92 ± 2.69 | 96.95 ± 2.24 | 96.90 ± 3.08 | 0.622 |
| Laboratory parameters | ||||
| RA | 5.87 ± 1.97 | 4.42 ± 0.69 | 7.32 ± 1.76 | <0.001 |
| RDW, % | 16.36 ± 2.64 | 14.99 ± 1.71 | 17.73 ± 2.69 | <0.001 |
| Albumin, g/dL | 2.97 ± 0.68 | 3.45 ± 0.49 | 2.50 ± 0.48 | <0.001 |
| WBC count, 109/L | 11.58 ± 16.15 | 10.96 ± 14.88 | 12.20 ± 17.29 | 0.026 |
| Hemoglobin, g/dl | 11.17 ± 2.03 | 11.75 ± 2.05 | 10.59 ± 1.84 | <0.001 |
| Hematocrit, % | 28.11 ± 6.12 | 29.89 ± 6.22 | 26.33 ± 5.47 | <0.001 |
| Platelet, 109/L | 201.01 ± 147.36 | 210.18 ± 136.92 | 191.89 ± 156.56 | <0.001 |
| Anion gap, mg/dl | 13.11 ± 3.78 | 13.11 ± 3.32 | 13.11 ± 4.18 | 0.979 |
| Bicarbonate, mg/dl | 24.52 ± 4.69 | 25.34 ± 4.24 | 23.71 ± 4.97 | <0.001 |
| Glucose, mmol/L | 140.50 ± 48.52 | 142.53 ± 45.72 | 138.48 ± 51.08 | 0.016 |
| Blood lactic acid, mmol/L | 3.46 ± 3.09 | 3.14 ± 2.83 | 3.73 ± 3.27 | <0.001 |
| Serum creatinine, mg/dl | 1.29 ± 1.32 | 1.24 ± 1.37 | 1.34 ± 1.27 | 0.019 |
| Serum urea nitrogen, mg/dl | 27.16 ± 22.25 | 24.13 ± 20.60 | 30.18 ± 23.40 | <0.001 |
| Serum sodium, mg/dl | 135.68 ± 5.59 | 135.97 ± 5.43 | 135.39 ± 5.73 | 0.003 |
| Serum potassium, mg/dl | 4.65 ± 0.93 | 4.60 ± 0.91 | 4.70 ± 0.94 | <0.001 |
| Severity of illness | ||||
| SOFA score | 5.18 ± 3.59 | 4.25 ± 3.06 | 6.10 ± 3.83 | <0.001 |
| Comorbidities | ||||
| CHF, n (%) | 442 (13.07) | 222 (13.17) | 220 (12.98) | 0.871 |
| CAD, n (%) | 501 (14.81) | 284 (16.84) | 217 (12.80) | <0.001 |
| Renal failure, n (%) | 440 (13.01) | 201 (11.92) | 239 (14.10) | <0.001 |
| Liver disease, n (%) | 388 (11.48) | 154 (9.13) | 234 (13.81) | <0.001 |
| Mortality, n (%) | ||||
| 30 days | 787 (23.28) | 265 (15.72) | 522 (30.80) | <0.001 |
| 90 days | 1131 (33.45) | 390 (23.13) | 741 (43.72) | <0.001 |
| 1 year | 1564 (46.26) | 608 (36.06) | 956 (56.40) | <0.001 |
| Length of ICU stay, day | 4.43 ± 6.14 | 3.90 ± 5.01 | 4.98 ± 7.07 | <0.001 |
Abbreviations: DBP—diastolic blood pressure; MAP—mean arterial pressure; RA—red blood cell distribution width‐to‐albumin; RDW—red cell distribution width; SBP—systolic blood pressure; SOFA—Sequential Organ Failure Assessment; WBC—white blood cell.
3.2. RA ratio is an independent risk factor for mortality in cancer patients
The relationship between RA ratio and patient mortality at 30 days, 90 days, and 1 year is outlined in Table 2. For 30 days mortality, the HR (95% CI) for the high RA ratio (≥5.51) was 2.17 [95CI% (1.87–2.51); p = <0.0001], compared with the low RA ratio (<5.51). In Model 2, sex and age were adjusted, and the HR (95% CI) for the high RA ratio (≥5.51) was 2.17 [95CI% (1.87–2.52); p < 0.0001], compared with the low RA ratio (<5.51). In Model 3, age, sex, anion gap, hematocrit, white blood cell count, congestive heart failure, SOFA, liver disease, and renal failure were adjusted, revealing HR (95% CI) for the high RA ratio (≥5.51) was 1.74 [95CI% (1.48–2.04); p < 0.0001], compared with the low RA ratio (<5.51). In the tertile groups, we found a significant association between higher RA ratios and 30 days all‐cause mortality (compared with the first dichotomized groups, <5.51) in model 1. The HR (95% CIs) for the three models for 30 days all‐cause mortality were 1.52 (1.23–1.87), and 2.18 (1.77–2.69), respectively (all p < 0.001). A similar relationship was also observed for 90 days and 1 year all‐cause mortalities.
TABLE 2.
HR (95% CIs) for all‐cause mortality of RA level
| Model 1 a | Model 2 b | Model 3 c | ||||
|---|---|---|---|---|---|---|
| HR (95% CIs) | p value | HR (95% CIs) | p value | HR (95% CIs) | p value | |
| 30‐Day all‐cause mortality | ||||||
| Dichotomized groups | ||||||
| <5.51 | 1 | 1 | 1 | |||
| ≥5.51 | 2.17 (1.87, 2.51) | <0.0001 | 2.17 (1.87, 2.52) | <0.0001 | 1.74 (1.48, 2.04) | <0.0001 |
| Tertile | ||||||
| <4.84 | 1 | 1 | 1 | |||
| 4.84–6.29 | 1.73 (1.41, 2.12) | <0.0001 | 1.70 (1.39, 2.08) | <0.0001 | 1.52 (1.23, 1.87) | <0.0001 |
| ≥6.29 | 2.99 (2.47, 3.61) | <0.0001 | 2.97 (2.46, 3.59) | <0.0001 | 2.18 (1.77, 2.69) | <0.0001 |
| 90‐Day all‐cause mortality | ||||||
| Dichotomized groups | ||||||
| <5.51 | 1 | 1 | 1 | |||
| ≥5.51 | 2.19 (1.93, 2.47) | <0.0001 | 2.19 (1.94, 2.48) | <0.0001 | 1.84 (1.61, 2.10) | <0.0001 |
| Tertile | ||||||
| <4.84 | 1 | 1 | 1 | |||
| 4.84–6.29 | 1.83 (1.55, 2.17) | <0.0001 | 1.81 (1.53, 2.14) | <0.0001 | 1.65 (1.39, 1.96) | <0.0001 |
| ≥6.29 | 3.03 (2.59, 3.55) | <0.0001 | 3.02 (2.58, 3.54) | <0.0001 | 2.40 (2.02, 2.85) | <0.0001 |
| One‐Year all‐cause mortality | ||||||
| Dichotomized groups | ||||||
| <5.51 | 1 | 1 | 1 | |||
| ≥5.51 | 1.92 (1.73, 2.12) | <0.0001 | 1.93 (1.74, 2.13) | <0.0001 | 1.65 (1.48, 1.84) | <0.0001 |
| Tertile | ||||||
| <4.84 | 1 | 1 | 1 | |||
| 4.84–6.29 | 1.64 (1.43, 1.87) | <0.0001 | 1.62 (1.42, 1.85) | <0.0001 | 1.47 (1.28, 1.69) | <0.0001 |
| ≥6.29 | 2.49 (2.19, 2.83) | <0.0001 | 2.48 (2.18, 2.82) | <0.0001 | 2.02 (1.75, 2.33) | <0.0001 |
Abbreviations: CI—confidence interval; HR—hazard ratio; RA—the ratio of RDW to albumin.
Models 1, 2, and 3 were derived from Cox proportional hazard regression models:
Model 1 covariates were adjusted for nothing.
Model 2 covariates were adjusted for age and sex.
Model 3 covariates were adjusted for age, sex, anion gap, hematocrit, white blood cell count, congestive heart failure, SOFA, liver disease, and renal failure.
Furthermore, linear regression was used to evaluate the association between RA ratio and length of stay, and obtained results were expressed as β (95% CIs). Results are outlined in Table 3. Results showed that the β (95% CIs) for the length of stay 1.08 (0.67, 1.50) 1.08 (0.67, 1.50), and 0.67 (0.23, 1.12) across the three models, respectively (all p < 0.001). Similar results were obtained in the tertile groups.
TABLE 3.
β (95% CIs) for length of ICU stay of RA ratio
| Model 1 a | Model 2 b | Model 3 c | ||||
|---|---|---|---|---|---|---|
| β (95% CIs) | p value | β (95% CIs) | p value | β (95% CIs) | p value | |
| Length of ICU stay | ||||||
| Dichotomized groups | ||||||
| <5.51 | 1 | 1 | 1 | |||
| ≥5.51 | 1.08 (0.67, 1.50) | <0.0001 | 1.08 (0.67, 1.50) | <0.0001 | 0.67 (0.23, 1.12) | 0.0032 |
| p for trend | <0.0001 | <0.0001 | 0.0032 | |||
| Tertile | ||||||
| <4.84 | 1 | 1 | 1 | |||
| 4.84–6.29 | 0.72 (0.21, 1.22) | 0.0054 | 0.73 (0.22, 1.24) | <0.0001 | 0.44 (−0.08, 0.97) | 0.0987 |
| ≥6.29 | 1.40 (0.90, 1.91) | <0.0001 | 1.41 (0.90, 1.91) | <0.0001 | 0.83 (0.26, 1.39) | 0.0042 |
| p for trend | <0.0001 | <0.0001 | 0.0047 | |||
Abbreviations: CI—confidence interval; RA—the ratio of RDW to albumin.
Models 1, 2, and 3 were derived from linear regression and used to evaluate the relationship between RA ratio and length of stay. Results were expressed as β (95% CIs).
Model 1 covariates were adjusted for nothing.
Model 2 covariates were adjusted for age and sex.
Model 3 covariates were adjusted for age, sex, anion gap, hematocrit, white blood cell count, congestive heart failure, SOFA, liver disease, and renal failure.
3.3. Subgroup analyses
Results from subgroup analyses are illustrated in Table 4. Summarily, we found no statistically significant association among factors in cancer patients.
TABLE 4.
Results from subgroup analysis showing the relationship between 30‐day all‐cause mortality and RA
| N | RA Ratio | p value | ||
|---|---|---|---|---|
| <5.51 | ≥5.51 | |||
| Clinical parameters | ||||
| Age, years | ||||
| ≤59 | 1127 | 1 | 2.59 (1.91, 3.51) | <0.0001 |
| 59–73 | 1125 | 1 | 2.46 (1.89, 3.19) | <0.0001 |
| ≥73 | 1129 | 1 | 1.78 (1.43, 2.23) | <0.0001 |
| Sex | ||||
| Female | 1356 | 1 | 2.17 (1.72, 2.73) | <0.0001 |
| Male | 2025 | 1 | 2.16 (1.79, 2.62) | <0.0001 |
| Vital signs | ||||
| Heart rate, beats/min | ||||
| ≤66 | 1074 | 1 | 2.80 (2.12, 3.69) | <0.0001 |
| 67–81 | 1167 | 1 | 2.08 (1.59, 2.71) | <0.0001 |
| ≥82 | 1134 | 1 | 1.66 (1.32, 2.10) | <0.0001 |
| SBP, mmHg | ||||
| ≤106 | 1125 | 1 | 2.38 (1.86, 3.04) | <0.0001 |
| 107–121 | 1123 | 1 | 1.73 (1.34, 2.25) | <0.0001 |
| ≥122 | 1126 | 1 | 1.59 (1.18, 2.14) | 0.0022 |
| MAP, mmHg | ||||
| ≤71 | 1127 | 1 | 2.45 (1.92, 3.13) | <0.0001 |
| 72–80 | 1123 | 1 | 1.95 (1.50, 2.53) | <0.0001 |
| ≥81 | 1124 | 1 | 1.45 (1.08, 1.94) | 0.0133 |
| Respiratory rate, bate/min | ||||
| ≤19 | 1680 | 1 | 2.53 (1.99, 3.22) | <0.0001 |
| ≥20 | 1686 | 1 | 1.84 (1.53, 2.22) | <0.0001 |
| SPO2, % | ||||
| ≤96.36 | 1126 | 1 | 2.22 (1.77, 2.79) | <0.0001 |
| 96.37–98.23 | 1120 | 1 | 2.60 (1.95, 3.47) | <0.0001 |
| ≥98.24 | 1125 | 1 | 1.87 (1.44, 2.45) | <0.0001 |
| Laboratory parameters | ||||
| WBC count, 109/L | ||||
| ≤6.5 | 1101 | 1 | 2.32 (1.75, 3.08) | <0.0001 |
| 6.6–11.5 | 1150 | 1 | 2.13 (1.63, 2.77) | <0.0001 |
| ≥11.6 | 1130 | 1 | 2.02 (1.60, 2.55) | <0.0001 |
| Hematocrit, % | ||||
| ≤25.2 | 1117 | 1 | 1.87 (1.40, 2.49) | <0.0001 |
| 25.3–30.2 | 1131 | 1 | 2.73 (2.10, 3.55) | <0.0001 |
| ≥30.3 | 1133 | 1 | 2.14 (1.66, 2.75) | <0.0001 |
| Hemoglobin, g/dL | ||||
| ≤8.4 | 1066 | 1 | 1.98 (1.46, 2.67) | <0.0001 |
| 8.5–10.1 | 1166 | 1 | 2.34 (1.81, 3.02) | <0.0001 |
| ≥ 10.2 | 1149 | 1 | 2.31 (1.79, 2.99) | <0.0001 |
| Platelet count, 109/L | ||||
| ≤120 | 1126 | 1 | 2.18 (1.69, 2.81) | <0.0001 |
| 121–233 | 1125 | 1 | 2.33 (1.78, 3.06) | <0.0001 |
| ≥234 | 1130 | 1 | 1.79 (1.39, 2.31) | <0.0001 |
| BUN, mg/dL | ||||
| ≤17 | 1002 | 1 | 2.03 (1.44, 2.87) | <0.0001 |
| 18–33 | 1248 | 1 | 1.88 (1.48, 2.41) | <0.0001 |
| ≥34 | 1131 | 1 | 1.97 (1.57, 2.46) | <0.0001 |
| Serum creatinine, mg/dL | ||||
| ≤0.6 | 828 | 1 | 1.83 (1.33, 2.52) | 0.0002 |
| 0.7–1.1 | 1419 | 1 | 2.37 (1.85, 3.04) | <0.0001 |
| ≥1.2 | 1134 | 1 | 1.97 (1.57, 2.47) | <0.0001 |
| Anion gap | ||||
| ≤10 | 753 | 1 | 2.43 (1.60, 3.69) | <0.0001 |
| 11–13 | 1333 | 1 | 2.02 (1.57, 2.59) | <0.0001 |
| ≥14 | 1291 | 1 | 2.37 (1.93, 2.91) | <0.0001 |
| Serum bicarbonate, mmol/L | ||||
| ≤19 | 1036 | 1 | 2.16 (1.66, 2.80) | <0.0001 |
| 20–23 | 1136 | 1 | 1.72 (1.32, 2.24) | <0.0001 |
| ≥24 | 1208 | 1 | 2.17 (1.68, 2.81) | <0.0001 |
| Direct bilirubin, μmol/L | ||||
| ≤0.4 | 814 | 1 | 2.63 (1.95, 3.55) | <0.0001 |
| 0.5–1.1 | 1094 | 1 | 1.99 (1.53, 2.57) | <0.0001 |
| ≥1.2 | 976 | 1 | 1.62 (1.23, 2.13) | 0.0006 |
| Serum chloride, mmol/L | ||||
| ≤99 | 1102 | 1 | 1.96 (1.58, 2.44) | <0.0001 |
| 100–104 | 1145 | 1 | 2.32 (1.76, 3.05) | <0.0001 |
| ≥105 | 1134 | 1 | 2.66 (1.96, 3.60) | <0.0001 |
| Serum glucose, mg/dL (min) | ||||
| ≤94 | 1126 | 1 | 1.99 (1.54, 2.58) | <0.0001 |
| 95–117 | 1089 | 1 | 2.15 (1.63, 2.82) | <0.0001 |
| ≥118 | 1166 | 1 | 2.29 (1.80, 2.92) | <0.0001 |
| Serum potassium, mmol/L | ||||
| ≤3.4 | 941 | 1 | 1.94 (1.45, 2.59) | <0.0001 |
| 3.5–3.9 | 1158 | 1 | 1.97 (1.52, 2.56) | <0.0001 |
| ≥4.0 | 1282 | 1 | 2.48 (1.97, 3.12) | <0.0001 |
| Serum sodium, mmol/L | ||||
| ≤133 | 947 | 1 | 1.73 (1.34, 2.23) | <0.0001 |
| 134–137 | 1170 | 1 | 2.07 (1.58, 2.70) | <0.0001 |
| ≥138 | 1264 | 1 | 2.50 (1.95, 3.21) | <0.0001 |
HRs (95% CIs) were derived from Cox proportional hazards regression models. Covariates were adjusted as in model 1 (Table 2).
3.4. Propensity score matching
Next, we performed propensity score matching to assess the relationship between RA ratio and cancer prognosis. Results are outlined in Table 5. We found no statistically significant differences among RA ratios of patients at baseline (Table 5). On the other hand, results from Cox regression analysis revealed that a high RA ratio was independently correlated to 30 days mortality (HR =1.33; 95% CI, 1.04–1.70; p = 0.0247).
TABLE 5.
Characteristics of the study population after propensity score matching
| Characteristics | RA Ratio | p value | |
|---|---|---|---|
| <5.51 | ≥5.51 | ||
| N | 555 | 555 | |
| Age, years | 65.70 ± 14.35 | 66.37 ± 13.69 | 0.4230 |
| Sex, n (%) | 0.8034 | ||
| Female | 207 (37.3) | 202 (36.4) | |
| Male | 348 (62.7) | 353 (63.6) | |
| Vital signs | |||
| SBP, mmHg | 116.40 ± 16.75 | 115.73 ± 17.26 | 0.5560 |
| DBP, mmHg | 61.17 ± 10.39 | 60.79 ± 11.27 | 0.5560 |
| MAP, mmHg | 77.08 ± 11.12 | 76.72 ± 11.89 | 0.6031 |
| Heart rate, beats/min | 91.53 ± 16.32 | 91.52 ± 16.39 | 0.9963 |
| SpO2, % | 97.07 ± 2.28 | 96.99 ± 2.51 | 0.5605 |
| Laboratory parameters | |||
| RA | 4.65 ± 0.61 | 7.01 ± 1.47 | <0.001 |
| RDW, % | 15.32 ± 1.75 | 17.39 ± 2.51 | <0.001 |
| Albumin, g/dL | 3.34 ± 0.49 | 2.54 ± 0.46 | <0.001 |
| WBC count, 109/L | 11.56 ± 18.80 | 11.85 ± 14.16 | 0.7256 |
| Hemoglobin, g/dl | 9.21 ± 1.76 | 9.25 ± 1.70 | 0.7552 |
| Hematocrit, % | 27.42 ± 5.44 | 27.51 ± 5.26 | 0.7791 |
| Platelet, 109/L | 192.11 ± 154.68 | 189.99 ± 149.62 | 0.8162 |
| Anion gap, mg/dl | 13.25 ± 3.87 | 12.91 ± 4.08 | 0.1579 |
| Bicarbonate, mg/dl | 20.96 ± 4.89 | 21.11 ± 5.35 | 0.4208 |
| Glucose, mmol/L | 144.37 ± 46.57 | 143.92 ± 58.28 | 0.8855 |
| Blood lactic acid, mmol/L | 1.85 ± 1.43 | 1.95 ± 1.50 | 0.2435 |
| Serum creatinine, mg/dl | 1.39 ± 1.22 | 1.39 ± 1.54 | 0.9588 |
| Serum urea nitrogen, mg/dl | 28.49 ± 24.27 | 28.98 ± 21.83 | 0.7256 |
| Serum sodium, mg/dl | 135.58 ± 5.49 | 135.58 ± 5.39 | 0.9340 |
| Serum potassium, mg/dl | 3.77 ± 0.60 | 3.77 ± 0.61 | 0.8741 |
| Severity of illness | |||
| SOFA score | 5.72 ± 3.34 | 5.89 ± 3.44 | 0.4208 |
| Comorbidities | |||
| CHF, n (%) | 89 (16) | 90 (16.2) | 1.0000 |
| CAD, n (%) | 96 (17.3) | 95 (17.1) | 1.0000 |
| Renal failure, n (%) | 86 (15.5) | 86 (15.5) | 1.0000 |
| Liver disease, n (%) | 82 (14.8) | 75 (13.5) | 0.6053 |
| Mortality, n (%) | |||
| 30 days | 113 (20.4) | 144 (25.9) | 0.0328 |
| 90 days | 163 (29.4) | 222 (40) | 0.0003 |
| 1 year | 228 (41.1) | 287 (51.7) | 0.0005 |
| Length of stay in ICU | 4.28 ± 4.84 | 5.02 ± 7.09 | 0.0420 |
Abbreviations: DBP—diastolic blood pressure; MAP—mean arterial pressure; RA—red blood cell distribution width‐to‐albumin; RDW—red cell distribution width; SBP—systolic blood pressure; SOFA—sequential organ failure assessment; WBC—white blood cell.
4. DISCUSSION
Systemic inflammation plays an important role in cancer progression. 28 , 29 Here, we provide the first report describing the RA ratio as an independent risk factor of all‐cause mortality in cancer patients with cancer. RDW is a classical indicator used to evaluate the size of circulating erythrocytes and assess the size variability, mainly in hematological, infectious, and cardiovascular diseases. 30 Although RDW significance in the early diagnosis of tumors has been revealed in recent years, its ability as a novel biomarker in early screening and prognostic evaluation remains unclear. RDW‐coefficient of variation is an independent indicator of colorectal cancer prognosis that can efficiently predict adverse recurrence and poor survival outcomes when combined with carcinoma embryonic antigen. Some studies have shown that high RDW may result from the overproduction of cytokines, such as TNF‐a and IL‐6, in the tumor microenvironment. 31 However, whether it is caused by systemic inflammatory response or poor chemotherapeutic effects remains unknown, necessitating further clarification. Elevated RDW, due to chronic inflammation, has been closely associated with erythrocyte deficiency, which is a part of the natural aging process in patients with cancer. In addition, elevated RDW in patients with cancer can result in anemia and poor nutrition status.
Albumin, a product of liver parenchymal cells that constitutes 40%–60% of the total plasma proteins, 32 is abundant in plasma where it is strongly associated with the nutritional status of the body. Previous studies have associated persistent systemic inflammatory responses with reduced albumin concentration in patients with advanced lung cancer or gastrointestinal tumors. In addition, inflammatory responses are reportedly stronger in youngers than middle‐aged and elderly patients. Notably, the occurrence of inflammation in the tumor microenvironment may not only initiate tumor development but also promote its progression.
The RA ratio may be a superior tool to other single identified markers in evaluating inflammatory response. In addition, it can serve as a prognostic marker owing to its ability to reflect tumor activities, thus can be used to identify high‐risk patients and as a therapeutic target to alleviate tumor progression. Since the RA ratio is rapidly and easily evaluated using laboratory examinations, it can function as a simple but relatively reliable index for the stratification of cancer patients at risk, even before ICU admission. To the best of our knowledge, this is the first report describing the relationship between the RA ratio and cancer survival outcomes. Notably, we used a large sample size, which increased the reliability of our findings.
This study had several limitations. Firstly, being a single‐center retrospective study, it may have been affected by selection bias, which potentially affected the accuracy of our results. Therefore, multicenter studies are needed to validate these findings. Secondly, we did not have a dynamic RA ratio in this study, and RDW was evaluated after ICU admission, which may have caused inevitable bias. Furthermore, the inclusion of more significant variables increases the predictive accuracy of a model. However, this study did not include some variables owing to missing data, which may have compromised model accuracy.
5. CONCLUSION
In summary, we provide the first evidence showing that increased RA is an independent predictor of increased all‐cause mortality in cancer patients. Additional prospective cohort studies are required to validate our findings.
CONFLICTS OF INTEREST
None.
ACKNOWLEDGMENTS
We thank Lihong Wang for help during the study.
Lu C, Long J, Liu H, et al. Red blood cell distribution width‐to‐albumin ratio is associated with all‐cause mortality in cancer patients. J Clin Lab Anal. 2022;36:e24423. doi: 10.1002/jcla.24423
Chengdong Lu and Jianyun Long contributed equally to this work.
Funding information
None
Contributor Information
Xin Fang, Email: hzfhfx@aliyun.com.
Yuandong Zhu, Email: 359089561@qq.com.
DATA AVAILABILITY STATEMENT
All the data used to support this study are available from the corresponding author upon request.
REFERENCES
- 1. Li J, Kang LN, Qiao YL. Review of the cervical cancer disease burden in mainland China. Asian Pac J Cancer Prev. 2011;12(5):1149. [PubMed] [Google Scholar]
- 2. Ferlay J, Shin HR, Bray F, Forman D, Mathers C, Parkin DM. Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008. Int J Cancer. 2010;127(12):2893‐2917. [DOI] [PubMed] [Google Scholar]
- 3. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer statistics. CA: A Cancer J Clin. 2011;6(2):169‐190. [DOI] [PubMed] [Google Scholar]
- 4. Santucci C, Carioli G, Bertuccio P, et al. Progress in cancer mortality, incidence, and survival: a global overview. Eur J Cancer Prev. 2020;29(5):367‐381. doi: 10.1097/CEJ.0000000000000594 [DOI] [PubMed] [Google Scholar]
- 5. Carioli G, Bertuccio P, Malvezzi M, et al. Cancer mortality predictions for 2019 in Latin. America. 2020;147(3):619‐632. [DOI] [PubMed] [Google Scholar]
- 6. Sopik V, Rosen B, Giannakeas V, Narod SA. Why have ovarian cancer mortality rates declined? part III. prospects for the future. Gynecol Oncol. 2015;138(3):757‐761. [DOI] [PubMed] [Google Scholar]
- 7. Lin S, Gao K, Gu S, et al. Worldwide trends in cervical cancer incidence and mortality, with predictions for the next 15 years. Cancer. 2021;127(21):4030‐4039. doi: 10.1002/cncr.33795 [DOI] [PubMed] [Google Scholar]
- 8. Wigmore T, Farquhar‐Smith P. Outcomes for critically ill cancer patients in the ICU: current trends and prediction. Int Anesthesiol Clin. 2016;54(4):e62‐e75. [DOI] [PubMed] [Google Scholar]
- 9. Takahashi S, Miura N, Harada T, et al. Prognostic impact of clinical course‐specific mRNA expression profiles in the serum of perioperative patients with esophageal cancer in the ICU: a case control study. J Transl Med. 2010;8:103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Yong W, Jiang YZ, Qian WH, Gao JXJPO. Prognostic role of NLR in urinary cancers: a meta‐analysis. PLoS One. 2014;9(3):e9207. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Wen Y, Yang J, Han X. Fibrinogen‐to‐albumin ratio is associated with all‐cause mortality in cancer patients. Int J Gen Med. 2021;14:4867‐4875. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Peng Y, Guan X, Wang J, Ma J. Red cell distribution width is correlated with all‐cause mortality of patients in the coronary care unit. J Int Med Res. 2020;48(7):300060520941317. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Alimehmeti I, Grabova S, Lekli E, et al. Red blood cell distribution width (RDW) association with ischemic stroke among adults younger than 55 years. J Neurol Sci. 2013;333:e250. [Google Scholar]
- 14. Erdem A, Ceylan U, Esen A, et al. Clinical usefulness of red cell distribution width to angiographic severity and coronary stent thrombosis. Int J Gen Med. 2016;9:319‐324. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Huda AQ, Karim MR, Mahmud MA, et al. Use of acute physiology and chronic health evaluation (APACHE)‐II and red cell distribution width (RDW) for assessment of mortality of patients with sepsis in ICU. Mymensingh Med J. 2017;26(3):585. [PubMed] [Google Scholar]
- 16. Seyhan EC, Özgül MA, Tutar N, Ömür I, Uysal A, Altın S. Red blood cell distribution and survival in patients with chronic obstructive pulmonary disease. COPD. 2013;10(4):416‐424. [DOI] [PubMed] [Google Scholar]
- 17. Research SJJoCM . Is red cell distribution width a novel biomarker of breast cancer activity? data from a pilot study. J Clin Med Res. 2013;5(2):121‐126. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Huang DP, Ma RM, Xiang YQJM. Utility of red cell distribution width as a prognostic factor in young breast cancer patients. Medicine. 2016;95(17):e3430. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Noha M, Reports SJMB. Three gold indicators for breast cancer prognosis: a case‐control study with ROC analysis for novel ratios related to CBC with (ALP and LDH). Mol Biol Rep. 2019;46(2):2013‐2027. [DOI] [PubMed] [Google Scholar]
- 20. Grimm G, Haslacher H, Kampitsch T, et al. Sex differences in the association between albumin and all‐cause and vascular mortality. Eur J Clin Invest. 2009;39(10):860‐865. [DOI] [PubMed] [Google Scholar]
- 21. Wang X, Wang J, Wu S, Ni Q, Chen P. Association between the neutrophil percentage‐to‐albumin ratio and outcomes in cardiac intensive care unit patients. Int J General Med. 2021;14:4933‐4943. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Zhao N, Hu W, Wu Z, et al. The red blood cell distribution width–albumin ratio: a promising predictor of mortality in stroke patients. Int J General Med. 2021;14:3737‐3747. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Long J, Xie X, Xu D, et al. Association between red blood cell distribution width‐to‐albumin ratio and prognosis of patients with aortic aneurysms. Int J General Med. 2021;14:6287‐6294. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Seo Y, Yu J, Park J, et al. Red cell distribution width/albumin ratio and 90‐day mortality after burn surgery. Burns & Trauma. 2022;10:tkab050. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Johnson AEW, Pollard TJ, Shen L, et al. MIMIC‐III, a freely accessible critical care database. Sci Data. 2016;3:160035. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Sun H, Que J, Peng Y, et al. The neutrophil‐lymphocyte ratio: a promising predictor of mortality in coronary care unit patients — A cohort study. Int Immunopharmacol. 2019;74:105692. [DOI] [PubMed] [Google Scholar]
- 27. Wang J, Zhou D, Dai Z, Li Xiaokun. Association between systemic immune‐inflammation index and diabetic depression. Clin Interv Aging. 2021;16:97‐105. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Abdalla T, Almanfalouti V, Effenberger K, et al. Evaluation of the hamburg‐glasgow classification in pancreatic cancer: preoperative staging by combining disseminated tumor load and systemic inflammation. Cancers. 2021;13:23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Wu X, Hou S, Liu H. Systemic immune inflammation index, ratio of lymphocytes to monocytes, lactate dehydrogenase and prognosis of diffuse large B‐cell lymphoma patients. World J Clin Cases. 2021;9(32):9825‐9834. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Wang B, Aihemaiti G, Cheng B, Li Xiaomei. Red blood cell distribution width is associated with all‐cause mortality in critically ill patients with cardiogenic shock. Med Sci Monit. 2019;25:7005‐7015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Miyamoto K, Inai K, Takeuchi D, Shinohara T, Nakanishi T. Relationships among red cell distribution width, anemia, and interleukin‐6 in adult congenital heart disease. Circ J. 2015;79(5):1100‐1106. [DOI] [PubMed] [Google Scholar]
- 32. Kuntip N, Japrung D, Pongprayoon PJB. How human serum albumin‐selective DNA aptamer binds to bovine and canine serum albumins. Biopolymers. 2021;112(3):e23421. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
All the data used to support this study are available from the corresponding author upon request.