Background:

Malnutrition is a severe but modifiable risk factor for cancers. However, the relationship between malnutrition and the survival of patients with brain metastases has not been fully revealed. We aimed to evaluate the prevalence of malnutrition and assess its prognostic value on patients with brain metastases

Methods:

We retrospectively recruited 2,633 patients with brain metastases between January 2014 and September 2020. Three malnutrition scores were used to evaluate patients’ malnutrition status at their first admission, including controlling nutritional status, the nutritional risk index, and the prognostic nutritional index. The association between malnutrition and overall survival (OS) was estimated.

Results:

The three malnutrition scores were associated with each other and with body mass index (BMI). Malnutrition assessed by any of the three scores was significantly associated with poor OS. All three malnutrition scores were better indicators than BMI, and adding malnutrition scores to the Graded Prognostic Assessment (GPA) scoring system could significantly improve the accuracy of prognosis prediction.

Conclusions:

Malnutrition monitoring using any of the three malnutrition scores on patients’ first admission could be a better survival indicator for patients with brain metastases compared with BMI alone.

Impact:

Malnutrition is a more significant indicator of survival stratification compared with BMI. Adding malnutrition to the GPA score system achieves better survival prediction.

Cancer is one of the leading causes of death in the world (1, 2). Brain metastases were developed in approximately 20% of patients with cancer and were the primary source of mortality (3). It is mainly associated with lung cancer (20%–56%), breast cancer (5%–20%), and melanoma (7%–16%; ref. 4). Although the primary cancer type could affect patients’ survival, the development of brain metastases often means poor prognosis with a median survival of 2 to 21 months (5). Some prognostic factors have been raised, such as age, Karnofsky performance status (KPS), extracranial metastases, and number of brain metastases, but most of these factors cannot be easily modified (6).

Malnutrition is a common somatic comorbidity with up to 70% prevalence in patients with cancer (7). It is modifiable but often ignored in cancer care and treatment (8). It could have severe consequences, including poor clinical outcomes and higher mortality (9, 10). Previous study has demonstrated that low body mass index (BMI) or weight loss was associated with worse survival outcome in patients with brain metastasis (11). However, with the obesity epidemic, identifying malnourished patients has become problematic because weight loss could be masked by adiposity and fluid retention, which may lead misclassification of the actual nutrition status of patients (8). Several scoring systems based on blood routine examination have been raised to better evaluate malnutrition than BMI alone (12–14).

To the best of our knowledge, the evidence on the relationship between malnutrition and survival of patients with brain metastases is limited. The current study aimed to investigate the prevalence and clinical associations of malnutrition and to explore whether it is an independent prognostic factor in patients with brain metastases.

Study population

A retrospective chart review of consecutive patients diagnosed with brain metastases between December 2013 and September 2020 was conducted using the electronic inpatient medical records of the West China Hospital, Sichuan University. This study was approved by the ethics committees of the West China Hospital and the written consent for patients included in the study was exempt by the ethics committees because the study only used retrospective observation data (No. 2022127). The study was conducted in accordance with STROBE Statement (Supplementary Materials).

Patients with brain metastases were identified by the definite diagnosis of secondary malignant neoplasm of brain based on the International Classification of Diseases, Tenth Revision (ICD-10) codes (C79.31) in their electronic inpatient medical records. Of 3,803 patients, 1,170 patients were excluded if they had missing information for BMI, laboratory data about nutritional parameters, or follow-up information. Eventually, a total of 2,633 patients were included in the current study.

Data collection

Information was collected when patients were first diagnosed with brain metastases. Demographic and clinical data were acquired from the electronic medical records, including patients baseline characteristics (i.e., age, gender, weight, and height), clinical presentation (i.e., primary tumor location, extracranial metastases status), imaging results (i.e., number of brain metastases), and laboratory data about nutritional parameters (i.e., serum albumin, cholesterol, and total lymphocyte count). KPS for each patient was also evaluated independently through patients’ electronic medical records by two oncologists.

Assessment of malnutrition

Body mass index (BMI) was calculated by the body mass (in kilograms) divided by the square of the body height (in meters). Each patient was divided to underweight (<18.5 kg/m2), normal weight (18.5 to 24.9 kg/m2), overweight (25.0 to 29.9 kg/m2), or obese (≥ 30 kg/m2) group according to their BMI. Three malnutrition screening tools were used to evaluate the malnutrition status of patients, including the Controlling Nutritional Status (CONUT) score (12), the Nutritional Risk Index (NRI) (14), and the Prognostic Nutritional Index (PNI) score (13). According to the original study, each scoring system's detailed calculation was summarized in Fig. 1A.

Figure 1.

The prevalence of malnutrition assessed by three malnutrition scores in patients with brain metastases. A, The details of the calculation of each malnutrition score. B, The Venn diagram showing the cumulative frequency and relationships between three malnutrition scores. C, Percentage of malnutrition by each scoring systems according to BMI subgroups.

Figure 1.

The prevalence of malnutrition assessed by three malnutrition scores in patients with brain metastases. A, The details of the calculation of each malnutrition score. B, The Venn diagram showing the cumulative frequency and relationships between three malnutrition scores. C, Percentage of malnutrition by each scoring systems according to BMI subgroups.

Close modal

CONUT score was developed by Ulíbarri and colleagues (12), and it is applied as a screening tools for automatic daily assessment of the nutrition status of all hospitalized patients. It takes routine laboratory data into account including serum albumin, total cholesterol level, and total lymphocyte count. Score of 0–1, 2–4, 5–8, and 9–12 are considered absent, mild, moderate, and severe malnutrition, respectively.

PNI score was developed by Buzby and colleagues (13), and it is firstly used as quantitative estimate of postoperative complications according to the baseline nutritional status. The formula for calculating PNI score is 10*serum albumin (g/dL) + 0.005*total lymphocyte count (mm3). A score larger than 38 is considered absent malnutrition, a score between 35 and 38 is considered moderate malnutrition, and a score less than 35 is considered severe malnutrition. Previous studies had shown the potential prognostic role of PNI score (15, 16).

Geriatric NRI score was developed by Bouillanne and colleagues, which is a simple and accurate scoring system to evaluate malnutrition status and predict the risk of morbidity and mortality in hospitalized older patients (14). The formula of NRI is 1.489*serum albumin (g/L) + 41.7 * ideal body weight. The ideal body weight was calculated using equations of Lorentz, that is, [height (cm) −100] − [(height (cm) − 150)/4] for men and [height (cm) – 100] − [(height (cm) − 150)/2.5] for women (17). If the current weight was exceeded the ideal body weight, the ratio of current weight/ideal weight was set as 1 (14, 17). A score equal or larger than 100 is considered absent malnutrition, a score between 97.5 and 99.99 is considered mild malnutrition, a score between 83.5 and 97.49 is considered moderate malnutrition, and a score less than 83.5 is considered as severe malnutrition.

Outcome and follow-up

The main outcome of this study was overall survival (OS). For each patient, follow-up time in days was calculated from the date of first diagnoses of brain metastases to death, loss, or end (i.e., September 30, 2020) of follow-up, whichever came first. The survival status and the date of death were acquired from the Household Registration Administration System databases in Sichuan Province.

Statistical analyses

Continuous data are presented as the mean value with SD, and categorical data are presented as the number of each group with their percentages. T test or χ2 tests were used to compare differences between groups. Venn diagrams were used to demonstrate the relationship between the three malnutritional scores. Spearman and Pearson's analyses were used to test the correlation of categorical and continuous malnutritional scores, respectively.

Kaplan–Meier curves were used to visualize the time-to-event data, and log-rank tests were applied to compare OS differences between different malnutrition groups. Multivariate models were constructed to evaluate the prognostic impact of malnutrition. Variables that were previously demonstrated as independent prognostic factors of brain metastases in the Graded Prognostic Assessment (GPA) scoring system (18) were adjusted, including age (continuous variable), KPS (<70,70–80, or 90–100), extracranial metastases (yes or no), and number of brain metastases (1, 2–3, or >3). Missing data were treated as an unknown category when fitting models. Because the proportional hazards assumption for each variable was satisfied according to the statistical and graphical diagnostics test based on Schoenfeld residuals (Supplementary Fig. S1), the associations between malnutritional scores and the OS of patients were estimated using HRs with 95% confidence intervals (CI) by applying Cox regression models.

Four indicators, including C-statistics, the area under the receiver operating characteristic (AUROC) curves, continuous net reclassification improvement (cNRI), and integrated discrimination improvement (IDI), were used to compare the accuracy and agreement of different models using R package timeROC, survcomp, survIDINRI. All analyses were performed with R 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria, RRID:SCR_001905). A two-sided P < 0.05 was considered statistically significant.

Data availability

The data generated in this study are available upon request from the corresponding author.

Population characteristics

Of the 3,803 patients diagnosed with brain metastases from January 2014 to September 2020 at the West China Hospital, Sichuan University, 2,633 patients were eventually selected according to the inclusion criteria (Supplementary Fig. S2). A total of 1,627 death cases were identified and the 1-year mortality rate was 24.32% (95% CI, 22.68%–26.03%). The comparison of baseline characteristics between included and excluded patients was shown in Supplementary Table S1. As present in Table 1, among patients enrolled in the current study, the most common primary tumor was lung cancer (2,027 patients, 77%). 542 (20.6%) patients had extracranial metastases, and 1,055 (40.1%) patients had more than three brain metastases. 1,842 (70%) patients had normal BMI, and 277 (10.5%) patients had underweight BMI. Supplementary Table S2 demonstrated the baseline characteristics of patients stratifying by each score. All three scores showed that the prevalence of malnutrition was higher in older patients or patients with lower KPS.

Table 1.

The baseline characteristics of included patients with brain metastases.

Overall
N 2,633 
age, years (mean [SD]) 56.36 (11.61) 
Gender, male (%) 1508 (57.3) 
KPS categories (%) 
 90–100 824 (31.3) 
 70–80 1504 (57.1) 
 <80 291 (11.1) 
 Missing 14 (0.5) 
Extracranial metastases, yes (%) 2054 (78.0) 
Primary cancer site (%) 
 Lung cancer 2,027 (77.0) 
 Nasopharyngeal carcinoma 135 (5.1) 
 Breast cancer 83 (3.2) 
 Others 388 (14.7) 
Number of brain metastases (%) 
 1 829 (31.5) 
 2 to 3 237 (9.0) 
 >3 1,055 (40.1) 
 Missing 512 (19.4) 
Albumin, g/L [mean (SD)] 40.31 (4.81) 
Lymphocyte count, 109/L [mean (SD)] 1.37 (0.59) 
Cholesterol, mmol/L [mean (SD)] 4.47 (1.00) 
Chemotherapy, yes (%) 1,921 (73.0) 
Craniotomy, yes (%) 213 (8.1) 
Targeted therapy, yes (%) 912 (34.6) 
Immunotherapy, yes (%) 86 (3.3) 
Radiotherapy, yes (%) 1,351 (51.3) 
Alcohol consumption history, yes (%) 619 (23.5) 
Smoking history (%) 
 Never 1,681 (63.8) 
 Current 284 (10.8) 
 Ever 668 (25.4) 
Diabetes, yes (%) 231 (8.8) 
Hypertension, yes (%) 419 (15.9) 
BMI categories (%) 
 Underweight 277 (10.5) 
 Normal 1,842 (70.0) 
 Overweight 474 (18.0) 
 Obese 40 (1.5) 
Overall
N 2,633 
age, years (mean [SD]) 56.36 (11.61) 
Gender, male (%) 1508 (57.3) 
KPS categories (%) 
 90–100 824 (31.3) 
 70–80 1504 (57.1) 
 <80 291 (11.1) 
 Missing 14 (0.5) 
Extracranial metastases, yes (%) 2054 (78.0) 
Primary cancer site (%) 
 Lung cancer 2,027 (77.0) 
 Nasopharyngeal carcinoma 135 (5.1) 
 Breast cancer 83 (3.2) 
 Others 388 (14.7) 
Number of brain metastases (%) 
 1 829 (31.5) 
 2 to 3 237 (9.0) 
 >3 1,055 (40.1) 
 Missing 512 (19.4) 
Albumin, g/L [mean (SD)] 40.31 (4.81) 
Lymphocyte count, 109/L [mean (SD)] 1.37 (0.59) 
Cholesterol, mmol/L [mean (SD)] 4.47 (1.00) 
Chemotherapy, yes (%) 1,921 (73.0) 
Craniotomy, yes (%) 213 (8.1) 
Targeted therapy, yes (%) 912 (34.6) 
Immunotherapy, yes (%) 86 (3.3) 
Radiotherapy, yes (%) 1,351 (51.3) 
Alcohol consumption history, yes (%) 619 (23.5) 
Smoking history (%) 
 Never 1,681 (63.8) 
 Current 284 (10.8) 
 Ever 668 (25.4) 
Diabetes, yes (%) 231 (8.8) 
Hypertension, yes (%) 419 (15.9) 
BMI categories (%) 
 Underweight 277 (10.5) 
 Normal 1,842 (70.0) 
 Overweight 474 (18.0) 
 Obese 40 (1.5) 

Prevalence and clinical association of malnutrition

A total of 1,984 (75.4%) of patients were assessed as any degree malnourished, and 963 (36.6%) of patients were assessed as moderate to severe malnutrition by at least one malnutrition score. The discrimination ability of each scoring system varied from 7.4% (PNI) to 63.4% (CONUT) for any degree of malnutrition and 7.4% (PNI) to 30.0% (NRI) for moderate to severe malnutrition (Fig. 1B).

Correlation analyses showed that malnutrition status assessed by the three scoring systems was correlated with BMI (Supplementary Fig. S3). As presented in Fig. 1C and Supplementary Table S3, moderate to severe malnutrition was most commonly observed in underweight patients (PNI: 52, 18.8%; NRI: 227, 81.9%; and CONUT: 52, 18.8%) while only NRI scores have the discrimination ability to identify moderate to severe malnutrition patients in obese groups (NRI: 3, 7.5%). The corresponding relationship between CONUT, PNI, NRI, and BMI categories were shown in Supplementary Fig. S4 using the Sankey diagram.

The prognostic role of malnutrition

Malnutrition scores grouped by either their original categories (Fig. 2AC) or moderate to severe malnutrition (Fig. 2DF) could strongly stratify the OS of patients with brain metastases. In multivariate analysis, both continuous and categorical malnutrition scores were significantly associated with OS after being adjusted for variables included in the GPA scores [HR and their 95% CI (severe vs. absent): CONUT: 3.19 (1.99–5.11); PNI: 1.89 (1.47–2.43); NRI: 2.15 (1.74–2.66), all P < 0.001; Table 2; Supplementary Table S4].

Figure 2.

Kaplan–Meier curves for OS. Malnutrition scores were categorized by their original classification (AC) and by moderate to severe malnutrition (DF).

Figure 2.

Kaplan–Meier curves for OS. Malnutrition scores were categorized by their original classification (AC) and by moderate to severe malnutrition (DF).

Close modal
Table 2.

The Cox proportional analyses of malnutrition scores with OS in patients with brain metastases.

Univariate analysisMultivariate analysis
HR (95% CI)PC indexHR (95% CI)PC index
CONUT, continuous(Per 1-point increment) 1.14 (1.11–1.17) <0.001 0.57 1.12 (1.10–1.16) <0.001 0.597 
CONUT, categorical (normal nutrition as reference)   0.556   0.589 
 Mild risk 1.31 (1.18–1.46) <0.001  1.29 (1.16–1.44) <0.001  
 Moderate risk 2.00 (1.68–2.38) <0.001  1.85 (1.55–2.20) <0.001  
 Severe risk 3.44 (2.15–5.51) <0.001  3.19 (1.99–5.11) <0.001  
PNI, continuous(Per 1-point increment) 0.95 (0.95–0.96) <0.001 0.587 0.96 (0.95–0.97) <0.001 0.603 
PNI, categorical (normal nutrition as reference)   0.522   0.578 
 Moderate risk 1.62 (1.30–2.03) <0.001  1.45 (1.16–1.81) 0.001  
 Severe risk 2.05 (1.60–2.64) <0.001  1.89 (1.47–2.43) <0.001  
NRI, continuous(Per 1-point increment) 0.97 (0.96–0.97) <0.001 0.585 0.97 (0.97–0.98)  0.601 
NRI, categorical (normal nutrition as reference)   0.574   0.599 
 Mild risk 1.33 (1.15–1.55) <0.001  1.29 (1.11–1.50) <0.001  
 Moderate risk 1.62 (1.45–1.81) <0.001  1.54 (1.37–1.72) <0.001  
 Severe risk 2.41 (1.95–2.97) <0.001  2.15 (1.74–2.66) <0.001  
Univariate analysisMultivariate analysis
HR (95% CI)PC indexHR (95% CI)PC index
CONUT, continuous(Per 1-point increment) 1.14 (1.11–1.17) <0.001 0.57 1.12 (1.10–1.16) <0.001 0.597 
CONUT, categorical (normal nutrition as reference)   0.556   0.589 
 Mild risk 1.31 (1.18–1.46) <0.001  1.29 (1.16–1.44) <0.001  
 Moderate risk 2.00 (1.68–2.38) <0.001  1.85 (1.55–2.20) <0.001  
 Severe risk 3.44 (2.15–5.51) <0.001  3.19 (1.99–5.11) <0.001  
PNI, continuous(Per 1-point increment) 0.95 (0.95–0.96) <0.001 0.587 0.96 (0.95–0.97) <0.001 0.603 
PNI, categorical (normal nutrition as reference)   0.522   0.578 
 Moderate risk 1.62 (1.30–2.03) <0.001  1.45 (1.16–1.81) 0.001  
 Severe risk 2.05 (1.60–2.64) <0.001  1.89 (1.47–2.43) <0.001  
NRI, continuous(Per 1-point increment) 0.97 (0.96–0.97) <0.001 0.585 0.97 (0.97–0.98)  0.601 
NRI, categorical (normal nutrition as reference)   0.574   0.599 
 Mild risk 1.33 (1.15–1.55) <0.001  1.29 (1.11–1.50) <0.001  
 Moderate risk 1.62 (1.45–1.81) <0.001  1.54 (1.37–1.72) <0.001  
 Severe risk 2.41 (1.95–2.97) <0.001  2.15 (1.74–2.66) <0.001  

In multivariate analyses, models were additionally adjusted for age, KPS, extracranial metastases, and number of brain metastases.

As shown in Supplementary Fig. S5, all three malnutrition scores had better prognostic prediction accuracy than BMI [5-year AUROC for CONUT: 0.61 (0.56–0.66); for PNI: 0.67 (0.62–0.71); for NRI 0.64 (0.59–0.68); for BMI: 0.56 (0.51–0.61)]. When models added additional malnutrition assessment to the GPA model, all models achieved better prediction accuracy in OS than the original model [5-year AUROC for CONUT+GPA: 0.69 (0.64–0.74); for PNI+GPA: 0.71 (0.67–0.76); NRI+GPA: 0.69 (0.65–0.74); GPA: 0.67 (0.63–0.72)]. Furthermore, we assessed and compared the model discriminability using C-index, cNRI, and IDI. When compared with BMI and GPA variables, all malnutrition assessments brought substantial improvements of prognostic prediction ability on the BMI or GPA variables (Table 3). When malnutrition scores were compared, PNI scores significantly outperformed the CONUT score in 5-year cNRI and IDI (difference in cNRI: 0.183; P < 0.001; difference in IDI: 0.023; P < 0.001) while other comparisons of malnutrition scores did not yield any significant difference.

Table 3.

Comparative analysis of the discrimination of malnutrition scores for OS in patients with brain metastases.

CONUT vs. BMIPNI vs. BMINRI vs. BMI
ComparisonDifferencePDifferencePDifferenceP
C index 0.049 <0.001 0.056 <0.001 0.035 <0.001 
5-year cNRI 0.145 <0.001 0.199 <0.001 0.22 <0.001 
5-year IDI 0.022 <0.001 0.044 <0.001 0.031 <0.001 
CONUT vs. BMIPNI vs. BMINRI vs. BMI
ComparisonDifferencePDifferencePDifferenceP
C index 0.049 <0.001 0.056 <0.001 0.035 <0.001 
5-year cNRI 0.145 <0.001 0.199 <0.001 0.22 <0.001 
5-year IDI 0.022 <0.001 0.044 <0.001 0.031 <0.001 
NRI vs. CONUTPNI vs. CONUTPNI vs. NRI
ComparisonDifferencePDifferencePDifferenceP
C index 0.036 0.476 0.002 0.354 0.002 0.341 
5-year cNRI 0.029 0.535 0.183 <0.001 0.066 0.297 
5-year IDI 0.013 0.059 0.023 <0.001 0.009 0.158 
NRI vs. CONUTPNI vs. CONUTPNI vs. NRI
ComparisonDifferencePDifferencePDifferenceP
C index 0.036 0.476 0.002 0.354 0.002 0.341 
5-year cNRI 0.029 0.535 0.183 <0.001 0.066 0.297 
5-year IDI 0.013 0.059 0.023 <0.001 0.009 0.158 
CONUT +GPA variables vs. GPA variablesPNI +GPA variables vs. GPA variablesNRI+GPA variables vs. GPA variables
ComparisonDifferencePDifferencePDifferenceP
C index 0.030 <0.001 0.035 <0.001 0.033 <0.001 
5-year cNRI 0.099 <0.001 0.204 <0.001 0.171 <0.001 
5-year IDI 0.021 <0.001 0.033 <0.001 0.024 <0.001 
CONUT +GPA variables vs. GPA variablesPNI +GPA variables vs. GPA variablesNRI+GPA variables vs. GPA variables
ComparisonDifferencePDifferencePDifferenceP
C index 0.030 <0.001 0.035 <0.001 0.033 <0.001 
5-year cNRI 0.099 <0.001 0.204 <0.001 0.171 <0.001 
5-year IDI 0.021 <0.001 0.033 <0.001 0.024 <0.001 

Subset analyses on malnutrition and treatment outcomes

Of the included patients, 1,921 patients (73%) patients underwent chemotherapy, 1,351 patients (51.3%) underwent radiotherapy, 912 patients (34.6%) patients underwent targeted therapy, and 86 (3.3%) patients underwent immunotherapy. All malnutrition categories showed significant association with OS in different treatment subgroups (Supplementary Figs. S6–S7). In multivariate analyses, all malnutrition scores were significantly associated with poor prognosis after chemotherapy, radiotherapy, targeted therapy, or immunotherapy, except that for categorical PNI, the associations were no longer significant for patients who underwent immunotherapy or targeted therapy (Table 4).

Table 4.

The Cox proportional analyses of malnutrition scores with OS in patients with brain metastases in each treatment groups.

ChemotherapyRadiotherapyTargeted therapyImmune checkpoint blockade
HR (95% CI)PHR (95% CI)PHR (95% CI)PHR (95% CI)P
PNI Categorical 1.49 (1.20–1.84) <0.001 1.58 (1.21–2.05) <0.001 1.29 (0.91–1.82) 0.15 2.34 (0.78–7.08) 0.13 
 Continuous 0.96 (0.95–0.97) <0.001 0.96 (0.95–0.97) <0.001 0.96 (0.94–0.97) <0.001 0.92 (0.88–0.97) 0.002 
NRI Categorical 1.43 (1.27–1.61) <0.001 1.50 (1.30–1.73) <0.001 1.49 (1.25–1.78) <0.001 2.99 (1.63–5.48) <0.001 
 Continuous 0.98 (0.97–0.98) <0.001 0.97 (0.97–0.98) <0.001 0.97 (0.96–0.98) <0.001 0.95 (0.91–0.99) 0.01 
CONUT Categorical 1.46 (1.21–1.76) <0.001 1.50 (1.19–1.90) <0.001 1.41 (1.04–1.90) 0.03 3.41 (1.39–8.37) 0.007 
 Continuous 1.11(1.08–1.15) <0.001 1.13 (1.09–1.17) <0.001 1.12 (1.07–1.17) <0.001 1.26 (1.08–1.48) 0.004 
ChemotherapyRadiotherapyTargeted therapyImmune checkpoint blockade
HR (95% CI)PHR (95% CI)PHR (95% CI)PHR (95% CI)P
PNI Categorical 1.49 (1.20–1.84) <0.001 1.58 (1.21–2.05) <0.001 1.29 (0.91–1.82) 0.15 2.34 (0.78–7.08) 0.13 
 Continuous 0.96 (0.95–0.97) <0.001 0.96 (0.95–0.97) <0.001 0.96 (0.94–0.97) <0.001 0.92 (0.88–0.97) 0.002 
NRI Categorical 1.43 (1.27–1.61) <0.001 1.50 (1.30–1.73) <0.001 1.49 (1.25–1.78) <0.001 2.99 (1.63–5.48) <0.001 
 Continuous 0.98 (0.97–0.98) <0.001 0.97 (0.97–0.98) <0.001 0.97 (0.96–0.98) <0.001 0.95 (0.91–0.99) 0.01 
CONUT Categorical 1.46 (1.21–1.76) <0.001 1.50 (1.19–1.90) <0.001 1.41 (1.04–1.90) 0.03 3.41 (1.39–8.37) 0.007 
 Continuous 1.11(1.08–1.15) <0.001 1.13 (1.09–1.17) <0.001 1.12 (1.07–1.17) <0.001 1.26 (1.08–1.48) 0.004 

Models were additionally adjusted for age, KPS, extracranial metastases, and number of brain metastases.

Because lung is the most common organs of cancer origins (n = 2,027, 77%), we additional tested whether patients with lung cancer and with other cancers have concordant findings. The results showed that the prognostic role of all malnutrition categories was consistent between lung cancer and other cancers (Supplementary Table S5).

In this large retrospective cohort of brain metastases of solid tumors, we assessed the prevalence of malnutrition through three widely used nutrition assessment systems. Although heterogeneities exist among the scoring systems, they were all significantly associated with worse OS. Adding malnutrition scores into GPA risk score could substantially improve its ability of prognostic prediction. The specific malnutrition score that should be added to the GPA risk score will need to be validated in an independent cohort.

Our results showed that malnutrition is a common problem in patients with brain metastases at their first admission. 75.4% of patients with brain metastases were characterized as having any degree of malnutrition, and 36.6% were characterized as moderate to severe malnutrition by at least one malnutrition assessment. For metastatic cancers, the prevalence of malnutrition using either PNI, CONUT, or NRI was often evaluated separately (16, 19, 20). Few of them evaluated the associations among the three scores with sufficient sample power at the same time. We found correlations within the three malnutrition scoring systems in our cohort.

Previous studies have identified an association of malnutrition assessment scores with poor OS in multiple cancers, including nasopharyngeal carcinoma, renal cell carcinoma, brain cancer, and colorectal cancer (15, 21–24). However, few have investigated the prognostic value of malnutrition on brain metastases. Because most systemic therapy cannot penetrate through the blood-brain barrier, the development of brain metastases often means marked alterations in the clinical course of patients (25). It is logical to infer that these unique characteristics of brain metastases might complicate the applicability of the previous conclusion on malnutrition and cancers to brain metastases. In addition, different from other disease, BMI might not be suitable for measure the malnutrition status of patients with cancer because adiposity and fluid retention may mask the potential loss of weight (8). Our studies showed that malnutrition assessed by any of the three malnutrition assessment scales was significantly related to worse survival in brain metastasis. Furthermore, all of them showed better discrimination and accuracy in predicting patient's OS compared with BMI. When treated as continuous variables, none of the three malnutrition scoring systems significantly outperformed others.

Zhang and colleagues found malnutrition scoring could improve the prognostic ability of the tumor–node–metastasis (TNM) classification system for all-cause mortality (24). According to EANO–ESMO Clinical Practice Guidelines for brain metastasis from solid tumors, different from most of the nonmetastatic cancers, for the majority of patients with brain metastasis, the goal of treatment is to prolong their survival with acceptable toxicity instead of cure (26). In patients with brain metastasis, TNM staging may not be specific enough to evaluate the patient's prognosis; instead, GPA risk scoring was more often applied (18). In the current study, we found that by integrating any of the three malnutrition scores into risk factors that constituted the GPA risk score, the overall model performance of survival prediction was elevated compared with the original GPA models.

It was worth noticing that malnutrition scores but not BMI could also be a good prognostic predictor after chemotherapy, radiotherapy, and even treatment. This was in accordance with previous studies on advanced cancers. Takemura and colleagues have shown CONUT score was significantly associated with progression-free survival and cancer-specific survival in advanced renal cell carcinoma treated with nivolumab (27). Guller and colleagues investigated 99 patients with advanced head and neck cancer treated with anti–PD-1 or anti–CTLA-4 antibodies and also found malnutrition assessed by NRI was related to worse post immunotherapy outcomes (22). Our study showed that all three malnutrition scores were strongly associated with OS for patients with brain metastases who received immunotherapy in both univariate and multivariate analyses. The potential mechanisms beneath it might be the metabolic competition in the tumor microenvironment (28). Tumor-imposed metabolic restriction on glucose utilization could inhibit the nutrients intake for T cells, while immune checkpoint blockade could inhibit mTOR activity and glycolysis enzymes. With sufficient nutrients, the hyporesponsiveness status of T cells might be reversed (29, 30).

To the first of our knowledge, it is the largest study focused on malnutrition in patients with brain metastases. All of the assessments of malnutrition we used were widely used clinically and obtained from simple body measurements and routine blood testing, which makes it easily promoted in clinical practice. Subset analyses based on different treatment groups further demonstrated that the prognostic role of malnutrition scores was consistent within patients who received different therapies.

Several limitations should also be noted. First, the cohort only recruited patients from a single institute and was restricted to only Chinese patients, so our results may not apply to people of other racial or ethnic backgrounds. Second, potential selection bias and residual confounding may exist because of the retrospective study design. Furthermore, we only assessed patients’ malnutrition status after the first diagnosis of brain metastases. Studies on the changes in malnutrition status and their association with cancer prognosis are warranted.

In conclusion, the current study comprehensively evaluated the prevalence of malnutrition in patients with brain metastasis using three objective malnutrition scores. Malnutrition assessment could allow physicians to identify patients with possible worse survival before and after treatment. Adding malnutrition status based on any of the three malnutritional score into the conventional GPA risk model could largely improve the prediction ability for patients’ OS. Therefore, malnutrition evaluation using any of the three malnutritional score at patients’ first admission might be important to accurate the expected prognosis and assistant the clinical and therapeutic decision-making.

No disclosures were reported.

The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Z. Liu: Data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. Y. Zhang: Conceptualization, resources, data curation, formal analysis, supervision, writing–original draft, writing–review and editing. Y. Pei: Resources, data curation, software, writing–original draft, writing–review and editing. Y. He: Resources, data curation, software, writing–review and editing. J. Yu: Resources, data curation, software, writing–review and editing. R. Zhang: Resources, data curation, software, writing–review and editing. J. Wang: Data curation, investigation, writing–review and editing. W. Chong: Writing–review and editing. Y. Hai: Writing–review and editing. X. Peng: Conceptualization, supervision, funding acquisition, project administration, writing–review and editing. F. Fang: Supervision, funding acquisition, validation, project administration, writing–review and editing.

The work was supported by National Key Research and Development Program of China (2021YFE0206600 to X.C. Peng), Chengdu International Science and Technology Cooperation Program (2022-GH03–00004-HZ to X.C. Peng), National Natural Science Foundation of China (81672386 to X.C. Peng and 82271364 to Y. Zhang), the Sichuan Province Science and Technology Support Program (2020YFS0276, 2022YFSY0012, 2021ZYCD011, 2021YFSY0008, 2021CDDZ-25, 2021CDZG-24 to X.C. Peng), the Technology Innovation Project of Chengdu Science and Technology Bureau (2019-YF05–00459-SN to X.C. Peng), the innovation team project of Affiliated Hospital of Clinical Medicine College of Chengdu University (CDFYCX202203 to Y. Zhang), and Postdoctoral research and Development Fund and Translational medicine fund of West China Hospital (2020HXBH119 and CGZH19002 to X.C. Peng).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Note: Supplementary data for this article are available at Cancer Epidemiology, Biomarkers & Prevention Online (http://cebp.aacrjournals.org/).

1.
Heron
M
,
Anderson
RN
.
Changes in the leading cause of death: recent patterns in heart disease and cancer mortality
.
NCHS Data Brief
2016
;
1
8
.
2.
Li
J
,
Liu
Y
,
Jiang
J
,
Peng
X
,
Hu
X
.
Effect of telehealth interventions on quality of life in cancer survivors: a systematic review and meta-analysis of randomized controlled trials
.
Int J Nurs Stud
2021
;
122
:
103970
.
3.
Sacks
P
,
Rahman
M
.
Epidemiology of brain metastases
.
Neurosurg Clin N Am
2020
;
31
:
481
8
.
4.
Achrol
AS
,
Rennert
RC
,
Anders
C
,
Soffietti
R
,
Ahluwalia
MS
,
Nayak
L
, et al
.
Brain metastases
.
Nat Rev Dis Prim
2019
;
5
:
5
.
5.
Lamba
N
,
Wen
PY
,
Aizer
AA
.
Epidemiology of brain metastases and leptomeningeal disease
.
Neuro Oncol
2021
;
23
:
1447
56
.
6.
Venur
VA
,
Ahluwalia
MS
.
Prognostic scores for brain metastasis patients: use in clinical practice and trial design
.
Chinese Clin Oncol
2015
;
4
:
1
7
.
7.
Silva FR de
M
,
de Oliveira
MGOA
,
Souza
ASR
,
Figueroa
JN
,
Santos
CS
.
Factors associated with malnutrition in hospitalized cancer patients: a cross-sectional study
.
Nutr J
2015
;
14
:
123
.
8.
Walsh
D
,
Szafranski
M
,
Aktas
A
,
Kadakia
KC
.
Malnutrition in cancer care: time to address the elephant in the room
.
J Oncol Pract
2019
;
15
:
357
9
.
9.
Shakersain
B
,
Santoni
G
,
Faxén-Irving
G
,
Rizzuto
D
,
Fratiglioni
L
,
Xu
W
, et al
.
Nutritional status and survival among old adults: an 11-year population-based longitudinal study
.
Eur J Clin Nutr
2016
;
70
:
320
5
.
10.
Zhang
X
,
Tang
T
,
Pang
L
,
Sharma
SV
,
Li
R
,
Nyitray
AG
, et al
.
Malnutrition and overall survival in older adults with cancer: a systematic review and meta-analysis
.
J Geriatr Oncol
2019
;
10
:
874
83
.
11.
Lareida
A
,
Terziev
R
,
Grossenbacher
B
,
Andratschke
N
,
Roth
P
,
Rohrmann
S
, et al
.
Underweight and weight loss are predictors of poor outcome in patients with brain metastasis
.
J Neurooncol
2019
;
145
:
339
47
.
12.
de Ulíbarri
JI
,
González-Madroño
A
,
de Villar
NGP
,
González
P
,
González
B
,
Mancha
A
, et al
.
CONUT : a tool for controlling nutritional status. First validation in a hospital population
.
Nutr Hosp
2005
;
20
:
38
45
.
13.
Onodera
T
,
Goseki
N
,
Kosaki
G
.
Prognostic nutritional index in gastrointestinal surgery of malnourished cancer patients
.
Nihon Geka Gakkai Zasshi
1984
;
85
:
1001
5
.
14.
Bouillanne
O
,
Morineau
G
,
Dupont
C
,
Coulombel
I
,
Vincent
JP
,
Nicolis
I
, et al
.
Geriatric nutritional risk index: a new index for evaluating at-risk elderly medical patients
.
Am J Clin Nutr
2005
;
82
:
777
83
.
15.
Tu
X
,
Ren
J
,
Zhao
Y
.
Prognostic value of prognostic nutritional index in nasopharyngeal carcinoma: a meta-analysis containing 4,511 patients
.
Oral Oncol
2020
;
110
:
104991
.
16.
Yasar
HA
,
Bir Yucel
K
,
Arslan
C
,
Ucar
G
,
Karakaya
S
,
Bilgin
B
, et al
.
The relationship between prognostic nutritional index and treatment response in patients with metastatic renal cell cancer
.
J Oncol Pharm Pract
2020
;
26
:
1110
6
.
17.
Minamisawa
M
,
Seidelmann
SB
,
Claggett
B
,
Hegde
SM
,
Shah
AM
,
Desai
AS
, et al
.
Impact of malnutrition using geriatric nutritional risk index in heart failure with preserved ejection fraction
.
JACC Hear Fail
2019
;
7
:
664
75
.
18.
Sperduto
PW
,
Kased
N
,
Roberge
D
,
Xu
Z
,
Shanley
R
,
Luo
X
, et al
.
Summary report on the graded prognostic assessment: an accurate and facile diagnosis-specific tool to estimate survival for patients with brain metastases
.
J Clin Oncol
2012
;
30
:
419
25
.
19.
Ucar
G
,
Ergun
Y
,
Acikgoz
Y
,
Uncu
D
.
The prognostic value of the prognostic nutritional index in patients with metastatic colorectal cancer
.
Asia Pac J Clin Oncol
2020
;
16
:
e179
84
.
20.
Ma
LX
,
Taylor
K
,
Espin-Garcia
O
,
Anconina
R
,
Suzuki
C
,
Allen
MJ
, et al
.
Prognostic significance of nutritional markers in metastatic gastric and esophageal adenocarcinoma
.
Cancer Med
2021
;
10
:
199
207
.
21.
Kim
SI
,
Kim
SJ
,
Kim
SJ
,
Cho
DS
.
Prognostic nutritional index and prognosis in renal cell carcinoma: a systematic review and meta-analysis
.
Urol Oncol
2021
;
39
:
623
30
.
22.
Guller
M
,
Herberg
M
,
Amin
N
,
Alkhatib
H
,
Maroun
C
,
Wu
E
, et al
.
Nutritional status as a predictive biomarker for immunotherapy outcomes in advanced head and neck cancer
.
Cancers
2021
;
13
:
5772
.
23.
Liao
CK
,
Chern
YJ
,
Hsu
YJ
,
Lin
YC
,
Yu
YL
,
Chiang
JM
, et al
.
The clinical utility of the geriatric nutritional risk index in predicting postoperative complications and long-term survival in elderly patients with colorectal cancer after curative surgery
.
Cancers
2021
;
13
:
5852
.
24.
Zhang
Q
,
Qian
L
,
Liu
T
,
Ding
J-S
,
Zhang
X
,
Song
M-M
, et al
.
Prevalence and prognostic value of malnutrition among elderly cancer patients using three scoring systems
.
Front Nutr
2021
;
8
:
1
10
.
25.
Weidle
UH
,
Niewohner
J
,
Tiefenthaler
G
.
The blood-brain barrier challenge for the treatment of brain cancer, secondary brain metastases, and neurological diseases
.
Cancer Genomics Proteomics
2015
;
12
:
167
78
.
26.
Le Rhun
E
,
Guckenberger
M
,
Smits
M
,
Dummer
R
,
Bachelot
T
,
Sahm
F
, et al
.
EANO–ESMO clinical practice guidelines for diagnosis, treatment and follow-up of patients with brain metastasis from solid tumors
.
Ann Oncol
2021
;
32
:
1332
47
.
27.
Takemura
K
,
Yuasa
T
,
Fujiwara
R
,
Ito
M
,
Suzuki
H
,
Yonese
J
, et al
.
Prognostic significance of the controlling nutritional status (CONUT) score in patients with advanced renal cell carcinoma treated with nivolumab after failure of prior tyrosine kinase inhibitors
.
J Urol
2020
;
204
:
1166
72
.
28.
Chang
CH
,
Qiu
J
,
O'Sullivan
D
,
Buck
MD
,
Noguchi
T
,
Curtis
JD
, et al
.
Metabolic competition in the tumor microenvironment is a driver of cancer progression
.
Cell
2015
;
162
:
1229
41
.
29.
Kedia-Mehta
N
,
Finlay
DK
.
Competition for nutrients and its role in controlling immune responses
.
Nat Commun
2019
;
10
:
2123
.
30.
Cohen
S
,
Danzaki
K
,
MacIver
NJ
.
Nutritional effects on T-cell immunometabolism
.
Eur J Immunol
2017
;
47
:
225
35
.

Supplementary data