Prognostic impact of maximum standardized uptake value (SUVmax) of the primary lesion on survival in advanced non-small-cell lung cancer: A retrospective study

Background: Positron emission tomography/computed tomography (PET/CT) has been widely recognized for diagnosing and staging lung cancer, but the prognostic value of standardized uptake value (SUV) in patients with advanced non-small-cell lung cancer (NSCLC) remains controversial. We aim to explore whether uorodeoxyglucose uptake has prognostic signicance in patients with advanced NSCLC. Methods: We performed a retrospective analysis of the data of patients with advanced NSCLC who had undergone PET/CT before systemic treatment between June 2012 and June 2016. The relationship between the SUV of the pulmonary lesion and lesion size was evaluated via Spearman’s correlation analysis. We collected patients’ clinical and pathological data. Univariate and multivariate analyses were performed to analyze the factors inuencing survival. Results: Altogether, 157 patients with advanced NSCLC were included. Among these, 135 died, 13 survived, and 9 had incomplete data (median follow-up period of 69 months). SUV was correlated with lesion size and was signicantly greater for tumors ≥ 3 cm than for tumors <3 cm (10.2±5.4 vs. 5.6±3.3, t=-6.709, p=0.000). Univariate analysis showed that survival was associated with gender, tumor size, epidermal growth factor receptor (EGFR) gene mutation, SUV of the primary lung lesion, and treatment line. Multivariate analysis showed a signicant correlation between SUV of the primary lung lesion and survival. The mortality risk of patients with SUV ≤ 6 was 35% lower than that of patients with SUV >6 (95% CI 0.436-0.972, Wald value 4.400, p=0.036). Conclusions: The SUV of the primary lung lesion on PET/CT is signicantly correlated with survival in previously untreated patients newly diagnosed with advanced NSCLC.

further study. Thus, in this study, we aim to explore the prognostic value of FDG PET/CT in patients with advanced NSCLC. patients were included in the study. The patient selection owchart is shown in Fig. 1. The study was approved by Ethics Committee of Peking Union Medical College Hospital. We collected data on the patients' demographic characteristics, including sex and age. Clinical data, such as the patients' smoking history, weight loss, performance status (PS) score, tumor/node/metastasis (TNM) stage, number, and location of metastatic lesions, treatment line, and follow-up, were also collected. We also recorded the histopathologic characteristics of lung cancer, including their gene signatures, histology, and the presence of epidermal growth factor receptor (EGFR) mutations. Lastly, we considered the patients' PET/CT data, such as their lung lesion size as well as the SUVmax of the primary lung lesion and metastatic lesions. Patients were followed up for a median of 69 months (range, 2-83 months). The last follow-up was in June 2018.
Overall survival was de ned as the time between diagnosis and the last follow-up or death. The median age at diagnosis was 65 years, which was used as a boundary value to divide the patients into two groups. Staging was performed according to the International Association for the Study of Lung Cancer (IASCL) TNM staging system, 7th edition. Loss of weight was de ned as a 10% decline in weight within the past 6 months. The size of the primary lung lesion was determined according to the maximum diameter of the lesion. With respect to SUVmax, both the primary lung lesion and the metastatic lesions were included. Brain metastases were not counted, as PET/CT has a low sensitivity for brain metastases. For example, in this study, 33 patients had brain metastases, but only 14 of them were diagnosed via PET/CT. Treatment was divided into two types: rst-line treatment and second-line treatment or above. Because EGFR mutation testing, but not anaplastic lymphoma kinase (ALK) and c-ros oncogene 1 receptor tyrosine kinase (ROS1) gene testing, was already routinely performed during the study period, only the EGFR gene mutation was evaluated. In total, 51 patients with positive EGFR gene mutation received rst-line EGFR tyrosine kinase inhibitor (TKI) treatment, 106 patients received platinum-containing combination chemotherapy, and 39 patients received treatment combined with bevacizumab, while 81 patients received second-line treatment after progression.

Data analysis
All statistical analyses were performed using IBM SPSS Statistics 25 (Armonk, New York, USA). An independent sample t-test was performed to explore the association between the SUVmax of the primary lung lesion or the highest SUVmax among all lesions and survival. Likewise, the differences in SUVmax between the primary lung lesion and metastatic lesion were compared using this method. The SUVs were categorized into the following ranges: 0-2, 2-4, 4-6, 6-8, 8-10, 10-12, 12-14, and > 14. Survival curves were drawn to de ne an optimal cut-off value, and SUVs were then divided into two groups according to the cut-off. Survival differences between these two groups were then evaluated using the Kaplan-Meier method. We also analyzed the survival impact of various factors, including sex, age, smoking status, loss of weight, performance status (PS) score, lung lesion size, smoking histology, EGFR mutation, TNM staging, number and location of metastatic lesions, treatment line, and SUVmax of the lung lesion.
Spearman's correlation analysis was used to investigate signi cant relationships between the SUVmax of the primary lung lesion and tumor size. Univariate survival analysis was performed using the Kaplan-Meier method, and comparison of survival between groups was performed using the log-rank test. Multivariate analysis was performed using the Cox proportional hazards model. Hazard ratios (HR) with 95% con dence intervals (CI) were equally calculated. A p value of < 0.05 was considered signi cant.

Correlation between patients' clinical characteristics and SUVmax
Altogether, 157 advanced NSCLC patients were included. Among these patients, 135(86%) died, 13 (8%) survived, and 9 had incomplete follow-up data. The correlation between the patients' clinical characteristics and the SUVmax of the primary lung lesions as well as the highest SUVmax among all lesions is shown in Table 1. The SUV of the primary lung lesions was signi cantly higher than that of the metastatic lesions (p 0.05) ( Table 2). Further, the SUVmax of the primary lung lesions could better re ect the severity of the tumor (p 0.05).  The SUVmax of the primary lung lesions was divided into eight groups as follows: 0-2, 2-4, 4-6, 6-8, 8-10, 10-12, 12-14, and > 14. According to the median survival time of the patients in each group (Table 3.2) and the survival curve ( Fig. 2.2), SUV = 6 was determined as the appropriate cut-off value. Patients whose primary lung lesions showed SUVmax > 6 had signi cantly lower median survival time than those showed SUVmax ≤ 6 (16.  The optimal SUVmax cut-off for both primary lung lesions and all lesions was 6. This cut-off could better distinguish patients and predict prognosis. The SUV of the primary lung lesions was used for further analysis.

Analysis of factors in uencing survival
Univariate analysis showed that survival was correlated with sex, size of the primary lung cancer lesion (≤ 3 cm vs. >3 cm), EGFR mutation, the SUVmax of the primary lung cancer lesion (≤ 6 vs. >6), and treatment line (Table 4). Speci cally, the survival time was signi cantly longer among female patients (p = 0.021), those with EGFR mutation-positive tumors (p = 0.000), size of the primary lung lesions < 3 cm (p = 0.006), SUVmax of the primary lung lesions ≤ 6 (p = 0.016), and second-line treatment or above (p = 0.001). However, survival was not correlated with age, loss of weight, smoking status, PS score, histology, TNM stage, metastatic lesions, and number of metastases (p > 0.05). The Spearman's correlation analysis showed an interaction between SUV of the pulmonary lesion and lesion size (R-value = 0.578, P < 0.001). Therefore, only four factors were included in Cox multivariate analysis: sex, EGFR mutation, SUV of the primary lung lesion, and treatment line. In multivariate analysis, survival was still correlated with sex (p = 0.026), EGFR mutation (p = 0.001), SUVmax of the primary lung cancer lesion (≤ 6 vs. >6) (p = 0.036), and treatment line (p = 0.029). Patients with SUVmax ≤ 6 had a mortality risk of 0.651 (95% CI: 0.436-0.972, Wald value, 4.400, p = 0.036) compared with those with SUVmax > 6.

Discussion
The evaluation of SUV is intuitive and convenient, but whether SUVmax provides additional prognostic value in addition to the TNM stage remains to be explored [12,13,21]. Previous studies showed that higher SUVmax was associated with worse overall survival, but these studies were mostly based on early stage lung cancer [4,6,18,22],and results have been con icting. Furthermore, some studie [12,21,23]reported that SUV has no prognostic value independent of the TNM stage in advanced lung cancer. Hoang et al [23]. suggested that SUV cannot predict prognosis in patients with advanced lung cancer (stage IIIB and stage IV) because a continuous increase in SUV does not increase the risk of mortality. In contrast, Paesmans et al [21]. reported that SUV is an independent predictor for patients with stage I-III NSCLC but not for those with stage IV disease. The possible reason for this is that the median survival for stage IV patients is only 12 months. The short survival period resulted in failure to identify any independent prognostic value for SUVmax. Our study con rmed that SUV is a prognostic factor for overall survival in advanced lung cancer patients. Our analysis showed that patients with EGFR mutations accounted for 32.5% of the patients in our study, and the development of targeted therapy has resulted in signi cantly prolonged survival in patients with advanced NSCLC [24,25]. A high SUVmax indicates a worse severity of malignancy and more aggressive tumor proliferation [6,26]. Because of these aggressive biological behavioral characteristics, patients with a high SUVmax lose the chance to receive second-line treatment. Thus, the bene t from a variety of treatments and the overall survival in these patients is limited [27].Patients who received rst-line treatment had a signi cantly greater SUV than did those who received second-line treatment or more. The results appear to support this view. Studies on the optimal SUV threshold have also reported inconsistent ndings [2,4,6,13,15,19,28,29]. In this study, the median survival time and survival curves showed that an SUV of 6 was the optimal cut-off value. Meanwhile, Detterbeck et al. suggested that the correlation between SUV and prognosis is a gradual association without an absolute threshold [30].
With respect to the association between SUVmax and clinical factors, our study found that the highest SUVmax of all lesions signi cantly increased among patients with loss of weight, primary lung cancer lesion size > 3 cm, and number of metastases ≥ 2. The higher SUV may indicate more severe malignancy, shorter doubling time [6],and more aggressive invasion [17,26]. This would lead to a larger malignant lesion [2,4,8,11,14,26],more extensive organ involvement and more obvious weight loss [2,3,31].
In addition, we also found that the SUVmax of all lesions was markedly higher in patients with squamous cell carcinoma than in patients with non-squamous-cell carcinoma. Previous studies showed that SUV was correlated with histology, with the SUV of squamous cell carcinoma being greater than that of adenocarcinoma [2,8,10,11,21,32]. In contrast, we found no signi cant difference in the histology and SUV of primary lung cancer lesions. This may be because squamous carcinoma is a central lung cancer and thus lymph node enlargement is common and more likely to form in the cavity, mixed with obstructive pneumonia and atelectasis. Previous studies also showed that the SUV of poorly differentiated malignant lesions was higher than that of well-differentiated tumors [8,29]. We could not perform a more in-depth analysis in this study because the number of cases of squamous cell carcinoma was less than that of adenocarcinoma. Further studies with a well-balanced histology distribution are needed to evaluate the correlation between tumor histological types and SUV.
Our study has some limitations that should be considered. First, it is a retrospective study conducted only in a single center. Therefore, further studies with a larger sample size and conducted in a multicenter setting are needed to con rm our ndings. Second, the estimated SUV of primary lung cancer lesions was in uenced by various factors; including the patient's blood glucose levels, imaging time, reconstruction method, maximum absorption value, and analysis by the same observer. However, it is impossible to guarantee that all patients will receive PET/CT imaging under the same conditions in the real world. Third, the inclusion of only patients who received treatment may have led to selection bias.

Conclusion
We found that pretreatment SUV of the primary lung lesion on PET/CT is signi cantly associated with prognosis in patients with advanced NSCLC. SUV of the primary lung lesions had a stronger relationship with survival than the highest SUVmax among all lesions. The optimal SUV cut-off was 6, and patients with SUVmax > 6 had a signi cantly worse prognosis than did those with an SUVmax ≤ 6. Thus, further prospective studies are warranted to con rm the feasibility of PET/CT imaging-guided treatments in patients with stage IV NSCLC.
Abbreviations ALK, anaplastic lymphoma kinase; CI, con dence intervals; EGFR, epidermal growth factor receptor; FDG, uorodeoxyglucose; HR, Hazard ratios; IASCL, International Association for the Study of Lung Cancer; NSCLC, non-small-cell lung cancer; PET/CT, positron emission tomography/computed tomography; PS, performance status; ROS1, c-ros oncogene 1 receptor tyrosine kinase; SUV, standardized uptake value; TKI, tyrosine kinase inhibitor; TMN, tumor/node/metastasis staging system Declarations Ethics approval and consent to participate The study was approved by Ethic Committee of Peking Union Medical College Hospital and concent for publication. All patients provided written informed consent for the collection of their clinical data.