In the present study, we constructed a nomogram model based on the metabolic features of PET/CT and the inflammatory indicators of the peripheral blood to predict the PFS of patients with inoperable LA-NSCLC who received concurrent chemoradiotherapy. The model had promising discrimination, calibration, and clinical applicability. To the best of our knowledge, this is the first prognostic model for patients with LA-NSCLC that integrates the two dimensions of tumor metabolism and host immune inflammation. The model can provide valuable information to clinicians for the early identification of subgroups of populations of LA-NSCLC with a poor prognosis following CCRT. This will effectively optimize the treatment strategy at an early stage and provide patients with appropriate personalized care and intervention management, thus enabling individualized and precise treatment and an improved prognosis.

The metabolic-volume parameter of 18F-FDG PET/CT has been demonstrated to be an independent prognostic factor in various kinds of tumors. Using PET/CT as a potential prognostic indicator has also attracted extensive attention from scholars [9,10,11]. SUVmax presents information only about individual volume pixels within the tumor but does not evaluate the volume or heterogeneity of the metabolically active lesions. In contrast to it, SUVmean is the mean value of SUV within the sketched ROI, which reflects the mean uptake of 18F-FDG within the ROI and represents a superior picture of the metabolic activity of the tumor [24]. TLG, on the other hand, reflects the total glycolytic rate of the active tumor tissue. It indirectly indicates the active degree of tumor cells and has been suggested to have an advantage over the other metabolic features of PET/CT in the prognostic assessment of patients [25,26,27]. In the study by Moon et al., 234 patients with stage IV lung adenocarcinoma who underwent PET/CT before chemotherapy were analyzed. The multivariate Cox proportional risk regression model showed that TLG was a significant independent predictor of the PFS and OS of patients [25]. For patients with advanced NSCLC who did not undergo surgery, Yıldırım et al. [26] analyzed 110 patients with advanced (stage IIIa-IV) NSCLC, all of whom received CCRT after PET/CT. A multifactorial Cox proportional risk regression model revealed that only low TLG (< 225.7) was an independent predictor of the OS of patients. The above study suggested that the greater the tumor burden is, the higher the total metabolic rate and the shorter the time of cell multiplication, all of which indicated that the prognosis of the patients is relatively poor.

SII is a newly developed systemic immune inflammatory index that uses neutrophil, lymphocyte, and platelet counts to quantify systemic inflammation. Compared to a single or a combination of two indicators [22, 28], SII provides a more comprehensive picture of the host’s immune status and inflammatory response, and its predictive value may be superior to that of LMR, NLR, and PLR. SII has been demonstrated to be a novel prognostic factor for a variety of malignancies, including NSCLC [29,30,31,32,33,34]. Guo et al. [31] conducted a retrospective study of 569 patients with NSCLC who underwent surgery. The result revealed that only SII was an independent prognostic factor for OS according to the multivariate analysis. Their findings indicated that SII is a promising prognostic factor with a better predictive value than NLR and PLR for NSCLC patients who were treated with surgery. Deng et al. retrospectively analyzed 203 NSCLC patients who were treated with first-line generation EGFR tyrosine kinase inhibitors and evaluated the prognostic value of SII, NLR, and PLR. The multivariate analysis showed that NLR, PLR, and SII were independent prognostic factors for PFS, while only SII was an independent prognostic factor for OS. This finding also indicates that SII has a relatively higher prognostic value [32]. A retrospective analysis of patients with NSCLC who were treated with nivolumab revealed that pretreatment SII was an independent predictor of PFS and OS [33]. For patients with LA-NSCLC receiving concurrent radiotherapy, a retrospective study including 332 patients revealed that a high pretreatment SII was significantly associated with a low treatment response. The pretreatment SII was an important independent predictor of OS. Patients with a low SII had a significantly longer median OS than patients with a high SII (30 months versus 10 months) [34]. The predictive role of SII as a comprehensive assessment index can be illustrated by the functions of platelets, neutrophils, and lymphocytes. Platelets can promote the angiogenesis and metastasis of tumors and protect cancer cells from antitumor immune responses [35]. Neutrophils can participate in the proliferation and metastasis of tumors by releasing inflammatory mediators such as neutrophil elastase and interleukins [36, 37]. Contrary to the functions of platelets and neutrophils, tumor-associated lymphocyte infiltration is generally indicative of a good prognosis of patients, as the immune response prevents the growth and metastasis of the tumor [38].

Previous studies have investigated the relationship between the metabolic features of PET/CT and blood inflammation indicators as well as their prognostic value in malignancies [39,40,41]. For example, a study of patients with colorectal cancer demonstrated that NLR and LMR correlated significantly with MTV and TLG [39]. Studies of patients with head and neck cancers have also revealed a significant positive correlation between NLR and MTV and TLG [40]. A retrospective study of 132 patients with NSCLC demonstrated that there was a significant positive correlation between NLR and PLR with MTV and LTG., high NLR (≥ 6.34), PLR (≥ 291.6), MTV (≥ 79.3), and LTG (≥ 674.6) were significantly associated with a poor prognosis [41]. Based on the above theories, PET/CT reflects the functional metabolism of tumor cells and can be used to evaluate the biological behavior of tumors, while systemic inflammatory immunomarkers can indicate the balance between pro- and antitumor activity. Their combined application for predictive analysis of patient prognosis not only reflects the systemic inflammatory response status of the patients but also represents the metabolic profile of the tumors.

In this study, we constructed a predictive nomogram model for PFS based on SII, SUVmean, and TLG, which were selected by univariate regression and LASSO. No such predictive models based on a combined indicator have been reported before. The model is highly accurate and clinically adaptive compared to a single indicator. The model exhibits superior predictive performance both in the internal test set and in an independent external test set. There are other predictive models (scores or biomarkers) for the prognosis of patients with NSCLC that have been reported in previous studies. Matteo et al. [42] constructed an immune metabolic prognostic index (IMPI) for patients with NSCLC treated with nivolumab based on MTV and SII. The results showed that IMPI was significantly associated with the prognosis of patients. The mOS of patients with low, intermediate, and high IMPI was 17.5 months, 9.4 months, and 3.2 months, respectively (p < 0.01). In another study of 149 patients with stage III-IV NSCLC receiving chemotherapy, the researchers constructed a scoring system (SUV-LMR score) based on SUVmax and LMR. They found that the SUV-LMR score was not only significantly associated with the treatment response but was also an independent predictor of PFS and OS [43]. In addition, studies have reported the prognostic value of NLR, SII and bone marrow-to-liver SUVmax ratios (BLRs) in patients with advanced NSCLC treated with chemotherapy or immunotherapy [44]. Compared to these models, our nomogram has great advantages. In the present study, we screened indicators based on three dimensions of clinical information, PET features and blood inflammation markers to construct a predictive nomogram model. The richness of its predictors is more sophisticated. In addition to SII, SUVmean, and TLG, the sex of the patient and CEA were also utilized for the construction of the nomogram, since they have been demonstrated to correlate with the prognosis of patients. Moreover, unlike previous studies, we did not simply group patients according to the cutoff values and then assign a corresponding score, ignoring the magnitude of the contribution of factors to the prediction. Our nomogram sufficiently examines the contribution of each factor to the prognosis and grants them appropriate weights in the calculation of the risk score, this results in a more accurate and individualized prognostic risk score for patients. Importantly, we verified the performance of the nomogram in both the internal and external test sets and found that the nomogram displayed promising identification, goodness-of-fit, discriminative power, and clinical effectiveness. Overall, the combination of SII with SUVmean and TLG optimizes the performance of the nomogram model even more, which may provide a quantitative and pragmatic predictive tool for risk stratification of patients undergoing CCRT. Meanwhile, their combination expands the new perspective of integrating 18F-FDG PET/CT images and hematologic inflammatory indicators.

Of course, there are still some limitations of this study. First, this is not a prospective study, and there are some biases in the collection of the data of patients, such as small sample size and short observation time, which may have an impact on the stability of the results. Second, this study only used the features of pretreatment PET/CT and the pretreatment blood inflammation indicators for the construction of the model, and the above indicators may change considerably during the treatment period. Whether there are more appropriate time points for evaluation needs to be further explored. In addition, definitive evidence is lacking for the correlation between the features of PET and the inflammatory response. Therefore, basic studies, as well as prospective controlled clinical trials with larger sample sizes, are needed to validate our results.