Prognostic Value of Novel CARWL Score in Stage IIIC Non-Small-Cell Lung Cancer Patients Undergoing Concurrent Chemoradiotherapy

Objectives We explored the prognostic utility of the unique combination of C-reactive-protein-to-albumin ratio (CAR) and significant weight loss (WL > 5%) over the preceding 6 months, namely, the CARWL score, in stage IIIC non-small-cell lung cancer (NSCLC) patients who underwent concurrent chemoradiotherapy (CCRT). Methods For each patient, the CAR was calculated using C-reactive protein and albumin measurements obtained on the first day of CCRT: CAR = C-reactive protein ÷ albumin. The availability of an ideal CAR cutoff that may categorize patients into two distinct progression-free (PFS) and overall survival (OS) outcomes was explored by employing receiver operating characteristic (ROC) curve analysis. Patients were additionally divided into two groups based on their status of significant WL according to the well-recognized Delphi criteria. Then, the CARWL score was created by combining all feasible combinations of the CAR and significant WL groupings. The potential links between pretreatment CARWL groups and the post-CCRT OS and PFS outcomes were determined as the primary and secondary endpoints. Results This retrospective cohort study comprised a total of 651 stage IIIC NSCLC patients. ROC curve analysis indicated that rounded 3.0 was the ideal CAR cutoff (area under the curve (AUC): 70.1%; sensitivity: 67.8%; specificity: 65.9%), which categorized the patients into CAR < 3.0 (N = 324) and CAR ≥ 3.0 (N = 327) groups. There were 308 (47.3%) and 343 (52.7%) patients without and with significant WL, respectively. The created CARWL groups were CARWL-0: CAR < 3.0 and WL ≤ 5.0%; CARWL-1: CAR < 3.0 and WL > 5.0%, or CAR ≥ 3.0 and WL ≤ 5.0%; and CARWL-2: CAR > 3.0 and WL > 5.0%. The Kaplan–Meier curves showed that the PFS (14.2 vs. 11.4 vs. 7.5 months; P < 0.001) and OS (37.3 vs. 23.6 vs. 12.8 months; P < 0.001) durations were gradually and significantly lowered from the CARWL-0 to CARWL-2 groups. The CARWL score's significant impacts on PFS and OS outcomes were found to be independent of the other variables in the multivariate analysis (P < 0.001, for each). Conclusions Our findings indicate that the novel CARWL score, which accounts for pretreatment CAR and significant WL during the preceding 6 months, can reliably stratify newly diagnosed stage IIIC NSCLC patients into three groups with significantly different PFS and OS after definitive CCRT.


Introduction
Nearly one-third of all non-small-cell lung cancer (NSCLC) patients are diagnosed with stage III disease [1].According to the 8 th edition of the American Joint Committee on Cancer (AJCC) staging system, stage IIIC NSCLC represents a patient group presenting with both locally (T 3-4 ) and regionally (N 3 ) advanced cancer (T 3-4 N 3 M 0 ), namely, the group with the highest risk of distant metastases (DM) and the lowest survival results, excluding metastatic (M1) disease.Concurrent chemoradiotherapy (CCRT) is the current standard of treatment for these patients, based on multiple randomized clinical trials and meta-analysis results indicating that it outperforms sequential therapy [2][3][4][5].Regardless, such patients' survival outcomes remain bleak even with aggressive CCRT, and other endeavors, such as escalating RT doses up to 74 Gy, have failed to enhance these outcomes [6].Adjuvant immunotherapy is another recommended standard for these patients, but its usage remains challenging in many countries due to health insurance restrictions [7].
Although high local-and distant relapse rates due to relative resistance to currently available chemotherapy agents and RT appear to be the principal causes of poor results after radical CCRT, the clinical outcomes of such individuals vary dramatically even after identical CCRT regimens.Te present TNM (tumor-node-metastasis) staging framework's sole focus on the original tumor's size and its local and regional expansion, without reference to tumor-and host-related biological variables, may explain such enormous discrepancies [8,9].Terefore, the introduction of new biological markers that may supplement the current TNM staging framework may aid in more advanced prognostic stratifcation and, presumably, tailored management of these patients.
Te carcinogenesis process initiates systemic infammation, the seventh hallmark of cancer, with a resultant increase in proinfammatory cytokines that stimulate the increased hepatic synthesis of C-reactive protein (CRP), a widely researched measure of infammation [10][11][12].Excessive infammatory cytokine release not only increases capillary permeability and hence albumin loss, but it also reduces hepatocytes' ability to synthesize albumin [13].As a result, higher CRP levels, and hence immunological activation, are associated with lower serum albumin levels: a clear sign of defective immunity, exacerbated infammation, and malnutrition [14].In this respect, increased CRP to albumin ratio (CAR) or its variant, Glasgow Prognostic Score (GPS), have previously been linked to a worse survival rate in NSCLC patients treated with various oncological strategies [15,16].
Patients' performance status, quality of life (QOL), responsiveness to therapy, and prognosis have all been shown to be signifcantly impacted by weight loss (WL) at presentation, which is present to some degree in up to 80% of all NSCLC patients [17,18].Te rates and intensity of treatment-related adverse symptoms such as early satiety, dysphagia, anorexia, nausea, vomiting, tiredness, esophagitis, diarrhea, and infections may also be increased by WL [19].Such side efects may reduce the tolerance of CCRT and, as a result, the therapeutic response and prognosis.Signifcant WL, defned as WL > 5% in the last 6 months, is acknowledged as one of the main predictors of cancercachexia, which is unquestionably related to a poor prognosis in locally advanced NSCLC patients [20].Furthermore, Evans et al. specifed the components of CAR, namely, the levels of CRP (>5 mg/L) and albumin (<3.2 g/dL), among the biochemistry criteria for cancer-cachexia defnition in the Washington consensus statement, bridging the gap between the WL and chronic infammation [21].
Based on the encouraging basic and clinical evidence for CAR and WL in NSCLC patients, we hypothesized that the combination of baseline values of these two parameters could serve as a novel comprehensive biomarker in predicting the prognosis of stage IIIC NSCLC patients undergoing defnitive pre-CCRT, by simultaneously providing the patient's immune, infammation, and nutritional status.However, the prognostic value of combining the pretreatment CAR measurements and WL over the past 6 months, also known as the CARWL score, has not been studied in stage IIIC NSCLC patients who underwent CCRT, except for a conference abstract presented by our team [22].Terefore, we planned to assess the prognostic signifcance of the novel CARWL score in IIIC NSCLC patients who received platinum-based defnitive CCRT in the retrospective setting.

Patient Population.
Patients in stage IIIC (AJCC 8 th ed.) who received CCRT with conventionally fractionated 60-66 Gy thoracic RT and at least one chemotherapy cycle concurrently between January 2010 and December 2020 were identifed using an institutional retrospective database search.Patients had to fulfll the following criteria to be eligible for the study: aged between 18 and 80 years, Eastern Cooperative Oncology Group (ECOG) performance score of 0-1, body mass index (BMI > 20 kg/m 2 ), pathological proof for NSCLC (adenocarcinoma (AC) or squamous-cell carcinoma (SCC)), stage IIIC disease by diagnostic computerized tomography (CT) and 18F-fuorodeoxyglucose positron emission tomography-CT (PET-CT), available pre-CCRT brain magnetic resonance imaging (MRI) scans, detailed chemotherapy and RT details, as well as complete blood count and biochemistry test results.Patients with established malignant pleural/pericardial efusion involved contralateral supraclavicular lymph nodes, a history of RT/ chemotherapy, and inadequate pulmonary, cardiac, renal, or hepatic functions were considered ineligible for the study.

Ethics, Consent, and
Permissions.Before any patient data were collected, the study design was approved by the Baskent University Medical Faculty's institutional review board under Project No. KA20/067.All patients provided written informed consent prior to the start of CCRT for the collection and analysis of blood samples and pathologic specimens as well as the publication of their results, either personally or through legally authorized representatives.

2
Canadian Respiratory Journal

Treatment Details.
We carried out all thoracic RT plans in line with our institutional care standards for newly diagnosed stage IIIC NSCLC patients, which dictate the use of coregistered diagnostic CTand PET-CT information.RT was delivered with megavoltage linear accelerators by utilizing the intensity-modulated RT (IMRT) technique.All target volume defnitions, total and per fractional dosage specifcations, normal tissue tolerance dose limits, and prescribed concomitant chemotherapies were the same as previously reported [23].Elective nodal irradiation was not permitted as an institutional care standard for such patients.During the RT course, all patients received 60-66 Gy RT in 30-33 fractions (2 Gy per fraction) and 1-3 cycles of cisplatin/ carboplatin plus one of docetaxel, paclitaxel, or vinorelbine combinations.

Assessment of C-Reactive Protein-to-Albumin Ratio and
Weight Loss.Te pretreatment CAR was calculated for each patient using CRP and albumin measures obtained on the frst day of CCRT: CAR � CRP/Albumin.Similarly, the percentage diference between the weight measures acquired on the frst day of CCRT and the weight stated by the patient for his/her measure 6 months before the start of CCRT was calculated.Signifcant WL was defned as WL > 5% in the previous 6 months, according to the Delphi criteria described by Fearon and colleagues [20].

Evaluation Treatment Response.
Following completion of CCRT, scheduled follow-up visits and patient evaluations were performed every three months for the frst two years and every six months or more frequently, if demanded, after that.For therapeutic response assessment, all patients were evaluated by complete blood count and biochemistry tests, PET-CT, or chest CT (following confrmation of metabolic complete response on PET-CT).Te European Organization for Research and Treatment of Cancer (EORTC)-1999 guidelines were applied for this purpose.Additional radiologic and nuclear medicine imaging methods were exclusively employed in situations where there was a clinical suspicion of metastasis or for restaging of locoregionally recurrent NSCLC.

Statistical Methods.
Te primary endpoint of this research was to analyze if the CARWL score groups could achieve a statistically meaningful diference in overall survival (OS: time from the frst day of CCRT to the date of death or the fnal visit).Te secondary endpoint was progression-free survival (PFS: time from the frst day of CCRT to the date of the frst observation of disease progression or death or the fnal visit).Categorical variables were presented using percentage frequency distributions.Medians and ranges were used to express quantitative variables.We investigated the correlations between distinct groups using chi-square or Student's t-tests and Spearman correlations as indicated.Receiver operating characteristic (ROC) curve analysis was used to search for an optimal CAR cutof that might divide the research population into two groups with substantially diferent OS and PFS fndings.Patients were further divided into two groups based on whether they had signifcant WL before the CCRT, which was defned as WL > 5% in the last 6 months according to the Delphi criteria [19].Kaplan-Meier estimates and log-rank tests were used to assess the potential impact of various risk variables on OS and PFS.Te likely interactions between these variables and survival outcomes were investigated using the multivariate Cox proportional hazard model, with any two-sided P < 0.05 values being considered signifcant.
For comparative subgroup analyses between three or more groups, the Bonferroni correction was utilized to reduce the chance-related false-positive results.
Before proceeding to comparative analyses, ROC curve analysis was used to search the presence of ideal CAR cutofs with that would interact with treatment results.Our search yielded signifcance at the cutofs of 3.02 (area under the curve (AUC: 67.4%; sensitivity: 66.5%; specifcity: 65.7%) and 2.96 (AUC: 70.1%; sensitivity: 67.8%; specifcity: 65.9%) for PFS and OS, respectively (Figure 1).Because both cutofs were remarkably close to 3.0, we decided to use this value as the common cutof point for both PFS and OS comparisons.As a result, the research participants were split into two CAR groups: Group 1: CAR < 3.0 (N � 324) and Group 2: CAR ≥ 3.0 (N � 327).Comparisons between the two CAR groups revealed that the CAR ≥ 3.0 patients had signifcantly shorter median PFS (8.6 vs. 14.1 months; P < 0.001) and OS (16.7 vs. 31.0months; P < 0.001) durations than their CAR < 3.0 counterparts (Table 2).Indicating that the outcomes were steadily worsening with a high CAR value over time, the actuarial 5-year and 8-year PFS and OS rates were likewise inferior in the CAR ≥ 3.0 patients' group (Table 2).
Results of the univariate analysis indicated that, except for the pretreatment CAR and WL groups, none of the other factors had a statistically signifcant impact on the PFS or OS outcomes of the study cohorts (Table 3).As a result, we divided patients into four groups based on their pretreatment CAR and WL: Group-1: CAR < 3.0 and WL ≤ 5.0%, Group-2: CAR < 3.0 and WL > 5.0%, Group-3: CAR ≥ 3.0 and WL ≤ 5.0%, and Group-4: CAR > 3.0 and WL > 5.0%, respectively.Nevertheless, because the PFS and OS outcomes of Groups 2 and 3 were statistically insignifcant, we merged them into one group and created the three-tiered CARWL score: CARWL-0: CAR < 3.0 and WL ≤ 5.0%; CARWL-1: CAR < 3.0 and WL > 5.0%, or CAR ≥ 3.0 and WL ≤ 5.0%; and CARWL-2: CAR > 3.0 and WL>5.0%(Table 4).As shown in Figure 2 and Table 2, results of Kaplan-Meier estimates showed that the PFS (14.2 vs. 11.4 vs. 7.5 months; P < 0.001) and OS (37.3 vs. 23.6 vs. 12.8 months; P < 0.001) outcomes were gradually and signifcantly lowering from CARWL-0 to CARWL-2 score groups, despite nearly equal distributions of baseline disease and patient characteristics between the groups.Te   Canadian Respiratory Journal results of the multivariate analysis indicated that the CARWL score's signifcant impacts on PFS and OS outcomes were independent of the other variables (Table 3), establishing the novel CARWL score as a robust immuneinfammation and nutritional score for stage IIIC NSCLC patients.Further investigation into the source of this disparity found that the earlier and much larger rates of DM in the higher CARWL score groups were the most likely explanation since locoregional control rate diferences were not statistically signifcant.Afrming this explanation, DM was detected in 157 (76.2%) of 206 CARWL-2 patients, with 116 (73.9%) of them manifesting in the frst 9 months of CCRT, whereas corresponding rates for CARWL-1 and CARWL-0 patients were 40.6% and 29.2%, respectively (P < 0.001).

Discussion
Stage IIIC NSCLC (T 3-4 N 3 M 0 disease) is the penultimate stage before oligometastatic (IVA) or extensively metastatic (IVB) cancer in the current TNM staging framework (AJCC 8 th ed).Survival outcomes following almost similar therapies, on the other hand, may difer dramatically between individuals, which might be connected to biological differences such as systemic immune and infammation status, as well as nutritional condition.In this respect, the present study represents a frst by investigating the prognostic usefulness of the combination of pre-CCRT CAR and signifcant WL (>5%) over the previous 6 months, namely, the CARWL score, in patients who underwent defnitive CCRT for stage IIIC NSCLC.Te two main fndings of this study were as follows: frst, the pre-CCRT CARWL score was able to stratify stage IIIC patients into three groups with substantially diferent PFS (P < 0.001) and OS (P < 0.001), confrming its prognostic usefulness.Second, whilst CARWL-1 patients had essentially identical outcomes to typical IIIC patients, PFS and OS results for CARWL-0 and CARWL-2 patients were comparable to stage II and IVB NSCLC patients, implying the need for adjustment of the staging and individual treatment algorithms..017for signifcance.Groups 1 vs. 2: P < 0.001.Groups 1 vs. 3: P < 0.001.Groups 2 vs. 3: P < 0.001.Note.CARWL-0: CAR < 3.0 and WL ≤ 5.0%; CARWL-1: CAR < 3.0 and WL > 5.0%, or CAR ≥ 3.0 and WL ≤ 5.0%; and CARWL-2: CAR > 3.0 and WL > 5.0%.PFS: progression-free survival; OS: overall survival; ECOG: Eastern Cooperative Oncology Group; SCC: squamous-cell carcinoma; AC: adenocarcinoma; T-stage: RT: radiotherapy; Gy: gray; tumor stage; CAR: C-reactive protein-to albumin ratio; WL: weight loss; CARWL: C-reactive protein-to albumin ratio and weight loss score.Systemic infammation, the seventh hallmark of cancer, is triggered by carcinogenic processes, with a commensurate surge in proinfammatory cytokine synthesis and release [24].As a result of this triggering action, cytokines like tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6) enhance hepatic CRP production, leading to increased systemic infammation and decreased albumin synthesis [12,24].Additionally, cytokine-induced increased capillary permeability further lowers the albumin levels due to leakage into the interstitial space [13].Elevated CRP and accompanying exacerbated infammatory responses are, thus, associated with lower serum albumin levels [25].CRP and albumin are not just the components of the immune-infammation index CAR, but they are also strong predictors of liver function, malnutrition, and eventual WL in cancer patients, which can lead to deadly cachexia [13].CAR or its GPS variant has already been related to worse survival outcomes in NSCLC patients [15,16].Similarly, the Washington consensus statement identifed CAR components CRP (>5 mg/L) and albumin (<3.2 g/dL) as biochemical indicators of cancer-cachexia [21].When these facts are combined with the principal determinant of cancercachexia defnition, namely, the signifcant WL in the previous 6 months, it appears that there is a strong link between CAR components, WL, persistent infammation, defective immunity, cancer-cachexia, and cancer progression, all of which result in poor treatment response and patient prognosis.Confrming this connection, we found that 343 (52.7%) and 327 (50.2%) patients had signifcant WL and high CAR levels, respectively.Furthermore, 206 (31.6%) of those patients had both high CAR and signifcant WL simultaneously, forming the current study's CARWL-2 cohort.Hence, in the lack of similarly designed NSCLC research, our study's results seem to establish the link between persistent infammation and resultant WL (likely the most critical phase in the cancer-cachexia cascade) and poor clinical outcomes.
Although we revealed that a high CAR value and signifcant WL before CCRT are two notable predictors of survival outcomes in stage IIIC patients, the most noteworthy fnding of our research was the demonstration of the novel CARWL score's unique capability to stratify these patients into three separate PFS and OS groups.Because the outcomes of the CARWL-0, CARWL-1, and CARWL-2 score groups were very similar to those reported for stage II, IIIC, and IVB patients, our fndings appear to substantiate the novel CARWL score as a reliable and potent prognostic biomarker for such patients, given the inability of the traditional TNM staging framework for such discrimination across patients in the same disease stage.Furthermore, in the absence of 5-year survivors, a median OS of only 12.8 months in CARWL-2 patients could clinically indicate infammation-induced chemoresistance and/or radioresistance of the primary tumor and occult systemic metastases that are beyond the detection capability of current staging tools, including the FDG-PET-CT [26,27], emphasizing the urgent need to revise standard treatment protocols for these patients.Confrming the presence of resistant occult metastases before the onset of CCRT, DM was detected in 157 (76.2%) of 206 CARWL-2 patients, with 116 (73.9%) of them manifesting in the frst 9 months of CCRT.Tis discovery may clinically imply the implementation of more sophisticated staging tools, such as liquid biopsy, which may aid in the more accurate staging of these patients who require more intensive systemic therapies while sparing the already metastatic patients from the unavailing toxicities of radical CCRT.Alternatively, our fndings could indicate that adding highly efective anti-infammatory agents, such as the cyclooxygenase-2 inhibitor celecoxib, to the treatment algorithm of CARWL-2 patients may improve their results, given Komaki and colleagues' impressive results in unresectable locally advanced NSCLC patients [28].
Our research has certain limitations.First, because it was a single-center retrospective cohort analysis, the fndings presented here are prone to negative consequences of sampling bias.Second, these fndings may not

Conclusions
Although confrmatory research is crucial to back up present discoveries, the results of this hypothesis-generating research revealed that the novel CARWL score, which integrates pretreatment CAR and signifcant WL over the previous 6 months, can reliably stratify newly diagnosed stage IIIC NSCLC patients into three groups with signifcantly diferent PFS and OS after defnitive CCRT.

Figure 1 :
Figure 1: Results of the receiver operating characteristic curve analysis showing the association between the pretreatment CARWL score groups and survival outcomes: (a) progression-free survival; (b) overall survival.

Table 1 :
Baseline patient and disease characteristics.

Table 3 :
Univariate and multivariate analysis outcomes.
[29]y to other stage IIIC NSCLC patients, such as those with an ECOG 2 performance score and/or receiving diferent RT or chemotherapy regimens, since we included only patients with an ECOG performance score of 0-1 who underwent indistinguishable staging and treatment processes.Tird, we did not run CARWL group-specifc correlations with other immune-infammation and cachexia indicators such as blood-borne platelets, neutrophils, monocytes, lymphocytes, and chemokines like interleukin-6, which could have disclosed the precise mechanisms underlying the novel CARWL score and the survival outcomes of stage IIIC NSCLC patients.And fourth, current discoveries should be interpreted with caution because, albeit statistically insignifcant, discrepancies in salvage therapies may have inadvertently benefted one CARWL group over the others.As a result, until the results of well-conducted large-scale research are available, our fndings should be viewed with caution and regarded as hypothesis generators rather than frm recommendations.Yet, when viewed through Clark's prognostic factor defnition[29], our fndings appear to mark the CRAWL score as a reputable, simple to achieve and calculate, factually quantifable, reproducible, and reasonably priced novel prognostic score for stage IIIC NSCLC patients undergoing CCRT.