Cancer patients frequently encounter nutritional and metabolic derangement, which carry prognostic significance[28, 29] and is often amenable to treatment. Therefore, mandatory screening and assessment are established in many clinical and health care settings, aiming to increase awareness and allow early recognition and treatment[4, 31]. Therefore, screening and evaluation tools are expected to be brief, timesaving and efficient, especially in the outpatient setting.
Among the nutrition screening tools validated and well-used in the past decades, referring to NRS2002, MUST and MST, there is no consensus on which one is more efficient and precise in cancer patients. Besides, in the newly established GLIM framework, it’s suggested that any validated screening tools can be used in the first step, while which tool is better in certain population or setting is uncertain. For cancer patients, MUST[9, 21], NRS2002[33, 34] and the short form of MNA (MNA-SF) have been used as the screening tool in literature, depending on the age of included patients and clinical setting. There are also studies using GLIM criteria to make nutrition assessment omitting the first step[16, 31].
As for nutrition assessment, PG-SGA is specially designed for and commonly used in cancer population, while the GLIM framework is an objective tool newly established in order to diagnose malnutrition and is initially proved to be valuable in cancer patients in predicting prognosis[9, 16, 33, 34].
Nutrition screening or nutrition risk screening aims to increase awareness of malnutrition, however, the abnormal nutrition screening results neither necessarily represent malnutrition, nor can be taken to design individualized nutrition pathways, with the exception that NRS2002 can be used to guide nutrition support. For patients at risk of nutrition, assessment should be done in order to establish the diagnosis of malnutrition and severity grading. The accordance between nutrition screening tools and GLIM criteria does not indicate the value as the screening tool in the first step.
In this study, we not only compared the accordance of difference screening tools with GLIM criteria, but also evaluated the performance of them in the first step of GLIM framework. Although MST (κ = 0.666) and NRS2002 (κ = 0.383) were better corelated with the GLIM criteria among ambulatory cancer patients than MUST (κ = 0.383), using MUST (κ = 0.614) as the screening tool in the first step of GLIM framework resulted in much better accordance with PG-SGA than using MST (κ = 0.504) or NRS2002 (κ = 0.363). Hence, MUST may be the better screening tool among ambulatory cancer patients or in outpatient setting.
Still, when the first step is omitted, GLIM criteria had better performance compared to PG-SGA than using any screening tools in the first step, including the MUST. It’s understandable that applying any screening tool in the first step decreased the SE as well as increased the SP compared with omitting the first step. Since GLIM criteria omitting the first step has only a “fair” SE (61.3% (54.7-67.5)) and a high enough SP (97.9% (95%CI 95.4-99.1)) in this study, the use of screening tools as the first step unsurprisingly decreased the consistency between GLIM framework and PG-SGA. Regarding the high incidence and adverse effects of malnutrition in cancer patients, along with the performance of GLIM criteria omitting first step compared to PG-SGA, it may be reasonable to assess nutritional status in every ambulatory cancer patient using GLIM criteria without nutrition screening, but whether it could bring clinical benefit to patients warrants prospective controlled clinical trials to prove.
Unlike MST and MUST which indicate risk of malnutrition, NRS2002 represents the risk related to nutrition, predicts clinical outcome and therefor guides nutrition support, which have been proved by multiple retrospective and prospective studies. Xu et al. reanalyzed data of a multi-center prospective cohort study, which recruited 1831 cases who received fitting requirement nutrition support therapy (support cohort) or glucose-electrolytes infusion (nonsupport cohort) respectively. Of the 827 (45.2%) cases who were NRS2002 positive, 391 (21.4%) cases were identified as malnourished by GLIM criteria. For the other 436 (23.8%) cases who were NRS2002 positive but GLIM negative, the rate of infection was significantly lower in the support cohort than in the nonsupport cohort (8.0% vs. 15.7%; p=0.011), which indicated that GLIM criteria neglected half of the patients who could benefit from nutritional support by decreasing the rate of infection complications. On the other hand, would patients who are NRS2002 negative but GLIM positive benefit from nutritional support? Part of patients with a NRS2002 score ≤3 may also gain positive clinical effect from nutrition support, though the proportion of patients who might benefit decreased as the score decreased. Will GLIM positive represent the part of the NRS200 negative patients who will benefit from nutritional support? In our study, NRS2002 negative and GLIM criteria positive patients took up 12.8% of participants, which makes the question warrant to answer. However, up to now, there is no literature answering this question, since NRS2002 is commonly used as the indicator of nutritional intervention and researchers usually chose NRS2002 as the screening tool in the first step of GLIM criteria[33, 34]. Real world study may help up to answer the question initially.
As discussed above, we recommend routine screening of nutrition risk by NRS2002, and routine nutrition evaluation by GLIM framework omitting the screening step, if it were not for workload. When conditions do not warrant routine nutritional assessment, MUST could be chosen as the screening tool in the first step of GLIM framework for the sake of decreasing evaluating burden as well as maintaining accuracy.
Our study has some strengths. As a prospective cross-sectional observational study, we recruited 562 patients in a short period of time, namely 3 weeks, and all data for MST, MUST, NRS2002, PG-SGA and GLIM were collected concurrently, which ensured that no data was omitted, and the conclusion drawn from the study is more reliable than retrospective study. Our study not only assessed the accordance of between different nutrition screening tools and GLIM criteria, but also evaluated the diagnostic capacity of them as the first step of GLIM framework. It’s recommended that any validated nutrition screening tools can be used in the first step of GLIM framework, but, to our knowledge, there have been no published study focusing on the best choice of screening tools. Moreover, body composition information such as FFMI or ASMI was usually unavailable in published articles and considered as important and warranted in validation of GLIM criteria. In the current study, we applied BIA in every participant and provided relatively precise evaluation of reduced muscle mass. As shown in our recently published study focusing on interrater reliability of the GLIM criteria using the same data, the appliance of body composition measurement improved the performance of GLIM framework compared to PG-SGA.
All the same, there are several limitations in our study. Since it was a single-center observational study, results should be interpreted with caution. Food intake reduction was assessed via self-report on general amount of food intake instead of an in-depth diet history for sake of promoting feasibility, thus the accuracy of food intake evaluation was impaired. Inflammation was assessed by increased CRP and current infection in this study. However, the detection rate of CRP was low in our study, while no patients with current infectious disease was admitted in our day oncology unit, so the evaluation of inflammation status is inadequate, which could lead to certain bias. Moreover, though recommended by a recent review, PG-SGA as the gold standard to validate the GLIM criteria is not well-accepted by all clinician and nutritionist. Comprehensive evaluation or clinical outcome such as survival situation or complications rates could be more valuable in the validation of GLIM criteria. We need to follow up the participants to collect survival data and information of adverse complications in order to reveal the value of GLIM criteria in predicting clinical outcomes and guiding nutrition support.