Maximum Standardized Uptake Value in Lymph Nodes Measured by 18F-FDG PET/CT Predicts the Outcome of Patients with Oesophagus Squamous Cell Carcinoma

Background: To investigate the prognostic value of 18 F–uorodeoxyglucose positron emission tomography/computed tomography ( 18 F-FDG PET/CT) in esophagus squamous cell carcinoma (ESCC), PET/CT imaging characteristics were explored in the present study. Methods: Baseline PET/CT and clinical characteristics were collected in 125 patients with ESCC treated with radical radiotherapy from 2007–2016. The maximum standardized uptake value (SUVmax) of the primary gross tumour (SUVmax-T) and metastatic lymph node (SUVmax-N) were separately measured using X-tile. Overall survival (OS) and progression free survival (PFS) were estimated according to the Kaplan–Meier method. A multivariate Cox model was used to establish the independent prognostic factors. Results: The gross tumours presented higher 18F-FDG uptake than normal tissues with a median SUVmax-T of 12.5. The OS and PFS did not show signicant differences between patients with SUVmax-T ≥ 12.5 and those with SUVmax-T <12.5 (P>0.05). However, patients with SUVmax-N ≥ 11 had a signicantly worse OS and PFS than those with SUVmax-N <11 (P<0.05). A weak correlation was observed in SUVmax-T and SUVmax-N. The OS and PFS of patients with PET-negative LNs was signicantly better than those with PET-positive LNs. However, the OS and PFS of patients with one or two PET-positive LNs were not signicantly better than those with more than two PET-positive LNs. In the univariate analysis, cT stage, positive or negative lymph nodes on the PET-CT image and the SUVmax-N were established as signicant prognostic factors for both OS and PFS. In multivariate analysis, SUVmax-N was proved to be an independent predictor for OS and PFS. Conclusions: SUVmax-N, but not SUVmax-T, is an independent prognostic indicator for patients with ESCC.


Background
Though progress has been made in evaluation and treatment strategies in recent decades, oesophagus carcinoma remains a lethal disease. In comparison with the consistently increasing occurrence of oesophagus adenocarcinoma in developed countries, oesophagus squamous cell carcinoma (ESCC) still predominates in Asians [1]. Apart from surgery, concurrent chemoradiotherapy (CRT) has been widely accepted as an alternative radical treatment option for patients with inoperable locally advanced or unresectable ESCC, including patients with cervical oesophageal tumours [2]. However, for patients treated with CRT, no precise staging evaluation for tumour size and lymph node metastasis, which are the most signi cant established prognostic factors, is available due to a lack of specimens [3]. With respect to clinical evaluations of the number of metastatic lymph nodes, even positron emission tomography/computed tomography (PET/CT) and ultrasonography endoscopy can only provide a rough estimation [4,5]. The underlying implication of lymph nodes has not been fully understood.
By revealing the signi cantly increased glucose metabolism of tumour cells over that of normal cells, PET/CT has played an important role in the diagnosis, staging and restaging of tumours after neoadjuvant therapies, delineating target volume in radiotherapy, evaluating therapeutic and predicting the prognosis of cancer [5]. A few studies have focused on the correlation of metabolic parameters such as the standard uptake values (SUVs) of gross tumours and survival outcomes in patients with ESCC but have reached controversial conclusions. A meta-analysis that included 10 studies concluded that a high maximum of standard uptake value (SUVmax) of the primary tumour predicted a poor overall survival with a hazard ratio of 1.86 (95%CI,1.53-2.27), but the cut-point to de ne a high SUV ranged widely in these studies (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15) [6]. A multi-centre prospective study reported that the SUV of the baseline gross tumour had a negative prognostic value in oesophageal cancer patients [7]. In contrast, few studies have concentrated on metastatic lymph nodes. Yap W.K. et al reported that a high nodal SUVmax predicted poor outcomes in ESCC patients treated with de nitive chemoradiotherapy in a cohort of 62 patients [8].
Moreover, the nodal SUVmax correlates with prognosis in patients with head and neck cancer [9]. Therefore, our aim is to explore and validate the prognostic value of the metabolic characteristics of lymph nodes in a large cohort.

Methods
The electronic medical records of patients with ESCC who were treated with radical radiotherapy or concurrent chemoradiotherapy and underwent a baseline PET/CT scan in the XXXX between 2007 and 2016 were retrospectively reviewed. The inclusion criteria were as follows: 1) histologically con rmed squamous cell cancer without clinical evidence of metastasis evaluated by a combination of physical examination, PET/CT and ultrasound of abdomen and neck (if available); 2) treatment with concurrent chemoradiotherapy (radiation dose ≥ 50 Gy) or radiotherapy alone (radiation dose ≥ 60 Gy) as radical primary treatment without endoscopic resection; 3) PET/CT scan within one month before beginning treatment; and 4) clinical records including a complete history, complete physical examination, complete PET/CT Pretreatment FDG/PET scans were performed for staging purposes. Patients were asked to fast for at least 6 hours before the examination and were injected intravenously with 18F-FDG (7.4 MBq/kg). Images were acquired approximately 60 min after the intravenous administration of the tracer. Whole-body PET/CT emission scans were obtained from the base of the skull to the midthigh. 18F-FDG PET/CT was performed using a Siemens biograph 16HR PET/CT (Knoxville, Tennessee, USA). The FDG-PET images were interpreted by an experienced nuclear medicine doctor and correlated with the computed tomography images. The maximum standardized uptake value (SUVmax) of the primary gross tumour (SUVmax-T) and metastatic lymph node (SUVmax-N) were separately measured. Lymph nodes with SUVmax > 2.5 were de ned as PET positive. The number of PET-positive nodes was also recorded.

Surveillance
The surveillance protocol consisted of follow-up clinic appointments (every 3 months during the rst 2 years, every 6 months during the third to fth years, and every 12 months thereafter). Routine follow-up examinations included a chest CT with contrast, oesophagus barium X and ultrasound examination of the neck and abdominal sites. Upper gastrointestinal endoscopy was performed every 6-12 months or when symptoms indicated recurrence.

Statistical analysis
Analyses were mainly performed using SPSS version 15.0 (SPSS Inc., Chicago, USA) and GraphPad Prism 7.0 (GraphPad Software, Inc., San Diego, USA). All P values are two-tailed, and P < 0.05 was considered statistically signi cant. The correlation was evaluated by Pearson's correlation coe cient. The SUVmax cut-point for survival analyses was determined by X-tile software 3.6.1 (Yale University School of Medicine, New Haven, USA) [10]. The overall survival (OS) time was calculated from the date that treatment began until death or loss to follow-up. Progression free survival (PFS) was de ned as recurrence or distant metastasis identi ed by imaging studies or endoscopy with histological proof and/or requiring clinical interventions. Survival curves were estimated according to the Kaplan-Meier method, and statistical comparisons were performed by log-rank tests. Univariate Cox regression analysis was performed for all prognostic factors with respect to overall survival (OS) and progression-free survival (PFS). Multivariate Cox regression analysis was used to determine the independent prognostic factors.

Characteristics of the Patients
A total of 125 patients who met the inclusion criteria were included in our study ( Table 1). The chemotherapy included platinum-based, Fluorouracil-based or Paclitaxel-based regimens. The majority of patients were male (86.4%), and the median age was 64 years (interquartile range 56-70). Nearly twothirds of the patients were current smokers or ex-smokers, and approximately half of the patients were frequent alcohol consumers. Remarkably, patients with a high SUVmax-N were more likely to have linear relatives who were diagnosed with malignant tumours than patients with a low SUVmax-N, suggesting a genetic basis for tumour generation. A total of 69.6% of the patients presented with cT3 or cT4 tumours, and 75.2% of the patients were recognized to have clinical regional lymph node metastasis. The characteristics of the patients are summarized in Table 1.   (Table 2), and the multivariate analysis showed that SUVmax-N remained as an independent predictor for OS and PFS. To provide a clinically useful tool to predict prognosis, we constructed a nomogram that integrated the SUVmax-N and several clinicopathological risk factors associated with progression-free survival. The T stage, number of PET-positive LNs and SUVmax-N were included in the prediction model (Fig. 4). The Cindex of the nomogram was 0.644.

Discussion
In the current study, we investigated the correlation of SUVmax-T, SUVmax-N and prognosis in ESCC. In current analysis, the SUVmax-T of patiens with PET-positive LNs were signi cantly higher than those with PET-negative LNs. SUVmax-T was associated with metabolic tumor burden. Larger tumor burden always promotes local/regional metastasis [11]. Our study did not identify SUVmax-T as an appropriate prognostic predictor. The result was consistent with Vatankulu's study where metastatic lymph node SUVmax had an effect in predicting survival whereas primary tumor SUVmax did not have an effect [12]. This may be due to the fact that SUVmax-T only represented a few pixels instead of the whole tumour. Studies showed that among PET biomarkers, metabolic tumor volume (MTV) and total lesion glycolysis (TLG) both re ect the metabolic tumor burden and are considered to be the strongest prognostic factors, even more so than tumoral maximal standardized uptake values (SUVmax) [13]. Some studies have shown that MTV or TLG, after taking the tumour size into consideration, are better prognostic predictors than SUVmax alone [14,15]. However, tumour volume was not routinely measured in most hospitals in China, making it less practical for clinical use. Moreover, in studies that found survival differences between patients with high or low tumour SUVmax, the cut-point to de ne a high SUVmax varied considerably [6]. The range of SUVmax-T varied from 4.5 to 15. Therefore, we believe that SUVmax-T is not a practical prognostic factor that oncologists should consider when making decisions.
Our results were in line with those of a previous report from Yap W K and his colleagues that found that patients with oesophageal cancer treated by de nitive chemoradiotherapy who had high nodal SUVmax (≥ 7) on the baseline PET had poor overall survival [16]. Moreover, similar results have been obtained in studies on head and neck carcinoma and gastric cancer [17]. We did not nd that the number of PETpositive lymph nodes correlated with survival; the number of PET-positive LNs was not a good re ection of the pathological metastasis of lymph nodes due to the swelling and fusion of nodes, poor sensitivity for recognizing small malignant tissues by PET and the inability of PET to distinguish in ammation from tumour [18]. In Li's analysis study, FDG PET/CT exhibited high speci city of 95.6%, but sensitivity was only 45.0% in diagnosing the cervical lymph node metastasis [19]. This may be one reason why the correlation of lymph node metastasis and prognosis was not signi cant. Also this was a retrospective analysis. The prognostic value of SUVmax-N should be validated in larger and prospective cohorts. Our study did not contain patients treated with radical oesophagectomy, which is a population that potentially has clinical differences from our patients. In general, patients who have undergone oesophagectomy would have an earlier tumour stage, fewer metastatic lymph nodes and better general conditions than the patients in our study.
In combination analysis, the C-index of the nomogram based on T stage, number of PET-positive LNs and SUVmax-N was 0.644. The predictive index was not satis ed to predict prognosis. In Lee's research, the combined interpretation of an SUVmax of more than 2.6 with iso-or low CT attenuation [area under the curve (AUC): 0.846] showed signi cantly better diagnostic performance for detecting malignant lymph nodes than SUVmax only (AUC: 0.791) and size (AUC: 0.693) in a receiver operating characteristic curve analysis [20]. Look at this way, combination of PET-CT parameters including SUVmax-T, SUVmax-N and CT attenuation may be better in clinical application.

Conclusions
In summary, SUVmax-N, but not SUVmax-T, is an independent prognostic indicator in patients with ESCC that may be applied in clinical practice.
Abbreviations LN: lymph node; SUVmax-N: the maximum standardized uptake value of metastatic lymph node; SUVmax-T: the maximum standardized uptake value of primary tumor; OS: overall survival; PFS: progression free survival; PFS: progression free survival Declarations Ethics approval and consent to participate Written informed consent was obtained from all patients included in the study. The study protocol was approved by the Ethics Committee of Fudan University Shanghai Cancer Center.

Consent for publication
Not applicable Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.