INTRODUCTION
More than 50% of patients with advanced cancer are diagnosed as having malnutrition with high nutritional requirement due to altered metabolic rate [1,2]. Moreover nearly 20% of cancer patients are dying of malnutrition which is induced from tumor or treatment induced malnutrition [3]. In addition, patients under malnutrition have high risk of poor prognosis and treatment outcomes [4,5].
There have been several attempts to evaluate the patients’ nutritional status before treatment and its usefulness as a prognostic marker was studied [6-8]. Among markers evaluated in previous studies, anemia [9], body mass index (BMI) [10], and prognostic nutritional index (PNI) [11] were markers which could be easily calculated from baseline laboratory exam. These markers also had strong correlation with patients’ nutritional status. In addition, their roles in predicting poor prognosis have been verified in different types of cancer [6,12-14].
For the past few decades, cytotoxic chemotherapy has been the treatment of choice for patients with advanced cancer. Because of the side effects, selecting a patient tolerable for conventional chemotherapy had become greatly important before initiation of treatment [15]. In the era of molecular target therapy, small molecules such as tyrosine kinase inhibitor (TKI) demonstrated promising outcome with minimal side effects compared to conventional treatment [16]. Main treatment option of cancer with target mutation was changed from conventional chemotherapy to small molecular agents suitable for genetic profiles of tumor [17]. Now days, many patients who have not been candidates for treatment due to poor nutritional status are receiving cancer treatment using targeted agents [18].
Although nutritional status is a known prognostic marker for survival and treatment outcome of patients treated with cytotoxic chemotherapy, there has been no study conducted in patients treated with small molecular agents. We investigated the prognostic value of nutritional status in non-small cell lung cancer (NSCLC) patients with activating epidermal growth factor receptor (EGFR) mutations treated with EGFR TKI.
METHODS
Study population
A total of 2012 NSCLC patients who were treated with gefitinib or erlotinib from July 2002 to September 2014 in Seoul National University Hospital (SNUH) were reviewed. Among them, patients with confirmed activating EGFR mutation (either exon 19 deletion, or exon 21 point mutation [L858R or L861Q] by either direct DNA sequencing method or peptide nucleic acid clamping method), available pretreatment laboratory values, and physical measurement information were included. Finally, 630 patients were enrolled for the evaluation. The protocol of this study was reviewed by the institutional review board of SNUH (SNUH IRB No. 1409-143-612) and conducted in accordance with the World Medical Association’s Declaration of Helsinki.
Data collection
Medical history including pathology reports, laboratory values, imaging data, and physical measurement values were collected from medical records of individual patients. Pretreatment complete blood cell count including lymphocyte proportion and serum albumin level was retrieved for each patient. Response to treatment was evaluated using computed tomography images and described by Response Evaluation Criteria in Solid Tumor criteria.
Three nutritional factors were defined as below. Anemia was categorized by serum hemoglobin level under 13 g/dL for males and 12 g/dL for females. BMI was categorized into three groups: BMI less than 18.5, 18.5 to 25, and more than 25. PNI was calculated by the following formula: 10 × serum albumin level (g/dL) × 0.005 × absolute lymphocyte count (number/mm2)
Death date of patients was collected from civil registry of South Korea, and progression-free survival (PFS) and overall survival (OS) were calculated.
Statistical analysis
Hazard ratio (HR) for each factor was calculated by Cox-regression proportional hazard model and median PFS was calculated by Kaplan-Meier analysis. Multivariable analysis was performed using factors satisfying p values less than 0.05 at univariable analysis. PFS was defined as the period from the date of starting EGFR TKI to the date of cancer progression or all-cause mortality. OS was defined as the period from the date of starting EGFR TKI to the date of all-cause mortality.
Scoring system were created by conventional categorization and applied to the each nutritional factor. Higher scores were applied to categories with shorter median PFS. Score 2 was applied to the category with BMI less than 18.5, score 1 for 18.5 to 25, score 0 for more than 25. Score 1 was also applied to patients with anemia, PNI less than 45, and score 0 for without. The sum of scores was used for survival analysis.
A p value less than 0.05 was used as the cutoff value for statistical significance. Stata version 12.1 (Stata Co., College Station, TX, USA) was used for the statistical analysis and GraphPad Prism 5 (GraphPad Software, La Jolla, CA, USA) was used for image production.
RESULTS
Characteristics of study population
Among the 630 patients, there were 236 male (37.5%) and 394 female (62.5%). Pathologic type was adenocarcinoma in 588 patients (93.3%), and 536 (85.1%) have 0 or 1 of Eastern Cooperative Oncology Group performance status (ECOG PS). Gefitinib was used as TKI in 573 patients (91.0%) and the other 57 (9.0%) used erlotinib. There were 372 patients of anemia (43.2%), 45 of BMI (7.1%) lower than 18.5, and 177 of PNI (28.1%) lower than 45 (Table 1).
Hazard ratio of each nutritional variables
In univariable analysis, the HR for PFS in anemia was 1.29 (95% confidence interval [CI], 1.05 to 1.58; p = 0.015), BMI ranged from 18.5 to 25 was 1.36 (95% CI, 1.04 to 1.78; p = 0.024), BMI lower than 18.5 was 1.98 (95% CI, 1.28 to 3.06; p = 0.002), and PNI lower than 45 was 1.57 (95% CI, 1.26 to 1.96; p < 0.001). Multivariable analysis was performed with other variables such as age at diagnosis, ECOG PS, and treatment line of TKI. In the analysis, BMI ranged from 18.5 to 25 (adjusted HR [aHR], 1.31; 95% CI, 1.00 to 1.71; p = 0.047), BMI lower than 18.5 (aHR, 1.62; 95% CI, 1.04 to 2.52; p = 0.033), and PNI lower than 45 (aHR, 1.48; 95% CI, 1.18 to 1.85; p = 0.001) were independent factors for poor PFS, but anemia (aHR, 1.17; 95% CI, 0.95 to 1.45; p = 0.132) lost its statistical significance.
Same approach was applied for the OS. The HR for death in anemia was 1.51 (95% CI, 1.19 to 1.93; p = 0.001), BMI lower than 18.5 was 2.52 (95% CI, 1.46 to 4.34; p = 0.001), and PNI lower than 45 was 1.90 (95% CI, 1.47 to 2.45; p < 0.001). In multivariable analysis, all of three factors were independently significant. Detailed outcomes are shown in Table 2, Fig. 3.
Same approach was applied for the OS. The HR for death in anemia was 1.51 (95% CI, 1.19 to 1.93; p = 0.001), BMI lower than 18.5 was 2.52 (95% CI, 1.46 to 4.34; p = 0.001), and PNI lower than 45 was 1.90 (95% CI, 1.47 to 2.45; p < 0.001). In multivariable analysis, all of three factors were independently significant. Detailed outcomes are shown in Table 2, Fig. 3.
Actuarial difference in progression-free survival and overall survival
In our study, all of three factors related to nutritional status have proved useful as prognostic factors for PFS and OS with EGFR TKI treatment. There were decreased median PFS of 2.3 months in anemia, 3.9 in BMI ranged from 18.5 to 25, 7.1 in BMI lower than 18.5, and 3.0 in PNI lower than 45. Similar tendency was observed for OS except the subjects with BMI lower than 18.5 (median OS, 26.7 months), as compared with BMI ranged from 18.5 to 25 (median OS, 25.6 months). Detailed data is described in Table 3, and Kaplan-Meier curves for each factor is plotted in Fig. 1.
Risk analysis using scoring system
Each factor was scored by conventional distribution. The sum of scores was considered as a surrogated marker of patients’ nutritional status and survival analysis was done. The subjects with score 0 was considered as the best nutritional status group, and the subjects with score 4 was considered as the poorest nutritional status group. The subjects with the higher score showed the shorter median PFS and OS as shown in Table 3, Fig. 2. In the same manner, the subjects with high score showed increasing tendency of HR compared to the subjects with score 0.
DISCUSSION
Pretreatment nutritional status is an important factor to be considered when planning cancer treatment. Anemia, low BMI, and low PNI are factors which can be easily calculated and represent patients’ nutritional status. Our study showed usefulness of these factors as prognostic markers in patients treated with TKI. Moreover, effectiveness of the simple scoring system representing personal nutritional status was suggested.
There are two important contributing factors that are related to malnutrition in cancer patients. One is increased metabolic rates and the other is reduced food intake due to cancer or treatment related symptoms. Increased energy consumption is augmented by cytokines produced by tumor and they activate false adaptive system in the brain and alter the homeostasis of energy storage. As a consequence, muscle wasting and lipolysis proceed and more calories are needed to prevent the catabolism. As the disease progress, patients are more likely to be functionally impaired and show intolerance to anticancer drug [19]. The other factor, reduced food intake, can be managed with active nutritional intervention, appetite stimulants, and antiemetics [20]. Sometimes, however, there are limitations of nutritional intervention in gastrointestinal cancer with anatomical obstruction or head and neck cancer with dysphagia. For these reasons, baseline nutritional status is important and malnutrition can be a sign of poor prognosis with less treatment options.
Pretreatment risk assessment is routinely performed by complete history taking including comorbidities and performance status. Among the indices, performance status graded by ECOG PS [21], developed in 1982 by ECOG, is the most commonly used for prognostic index and has been proved its clinical significance as a prognostic marker [22]. In addition to the ECOG PS, nutritional status is emerging as an important prognostic factor [23,24].
Previous data showed active oral nutritional intervention has positive effect in decreased morbidity [25] and improves quality of life [26]. Moreover, there is a report that nutritional intervention improves tolerance to chemotherapy [4]. Regarding the above information, patients with higher scores in our study should be considered as a high risk population of treatment failure, and active nutritional intervention should be considered to these patients.
Both anemia and low PNI have been demonstrated to be associated with higher risk of poor prognosis. Although no previous study was on NSCLC patients with BMI, there have been controversial results, in which obesity was reported as a risk factor for early cancer related mortality. But this can be explained by the unique pathophysiology of breast cancer in relation to sex hormones being produced more in adipose tissue of obese patients [27-29]. Overall malnutrition as a risk factor in NSCLC should be understood in other ways. In the concept of risk factor paradox introduced in chronic disease, obesity showed protective effect in wasting disease, including cancer [30]. As patients with other wasting disease, our patients with high BMI showed lower incidence of adverse effects than patients with low BMI. This can be explained by time differential of competing risks [14]. Applying the time discrepancy between the groups categorized by nutrition factors, paradoxical effects of overweight are expected to be more prominent in the era of target therapy, due to prolonged survival and elongated duration of the treatment.
In our study, all of the three factors, anemia, BMI, and PNI, were significant as prognostic factors for PFS with EGFR TKI and OS. And these results maintained with the adjustment of ECOG PS. This is consistent with a previous study that demonstrated early progression of disease in cachexic patients compared to non-cachexic patients [31]. Although PS is routinely used as a means of representing overall physical condition in cancer patients, there is a shortage of evaluating nutritional status before treatment.
There are several limitations in our study. First, our study population is retrospective and does not represent the whole population treated with targeted agents. Nonetheless, EGFR TKI is commonly used small molecules and our study population has been well controlled by validated mutation analysis. Second, the scoring system used in our study has not been verified previously. Therefore, its usefulness should be confirmed in future studies.
Now days, increased number of patients are receiving cancer treatment using targeted therapy. However, individual nutritional status has being overlooked due to minimal side effects of small molecular agents. Since our study revealed the importance of baseline nutritional status as a prognostic factor, pretreatment evaluation of nutritional status and adequate nutritional support should be emphasized even in the patients receiving targeted therapy.