Prognostic Value of Restaging F-18 Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography to Predict 3-Year Post-Recurrence Survival in Patients with Recurrent Gastric Cancer after Curative Resection

Abstract

Objective

The aim of this study was to investigate the prognostic value of the maximum standardized uptake value (SUV_max) measured while restaging with F-18 fluorodeoxyglucose (¹⁸F-FDG) positron emission tomography/computed tomography (PET/CT) to predict the 3-year post-recurrence survival (PRS) in patients with recurrent gastric cancer after curative surgical resection.

Materials and Methods

In total, 47 patients with recurrent gastric cancer after curative resection who underwent restaging with ¹⁸F-FDG PET/CT were included. For the semiquantitative analysis, SUV_max was measured over the visually discernable ¹⁸F-FDG-avid recurrent lesions. Cox proportional-hazards regression models were used to predict the 3-year PRS. Differences in 3-year PRS were assessed with the Kaplan–Meier analysis.

Results

Thirty-nine of the 47 patients (83%) expired within 3 years after recurrence in the median follow-up period of 30.3 months. In the multivariate analysis, SUV_max (p = 0.012), weight loss (p = 0.025), and neutrophil count (p = 0.006) were significant prognostic factors for 3-year PRS. The Kaplan–Meier curves demonstrated significantly poor 3-year PRS in patients with SUV_max > 5.1 than in those with SUV_max ≤ 5.1 (3-year PRS rate, 3.5% vs. 38.9%, p < 0.001).

Conclusion

High SUV_max on restaging with ¹⁸F-FDG PET/CT is a poor prognostic factor for 3-year PRS. It may strengthen the role of ¹⁸F-FDG PET/CT in further stratifying the prognosis of recurrent gastric cancer.

INTRODUCTION

Gastric cancer is the sixth most common malignancy and the second leading cause of cancer deaths worldwide (1). Although the 5 year-overall survival (OS) rate is 62–71% in patients treated via surgery, a significant proportion of patients develop recurrences following resection (2,3). Approximately 1/3 of patients (35–42%) still relapse after curative resection and adjuvant chemotherapy in Asian countries (3,4). Hence, identifying relevant risk factors for patients with recurrent gastric cancer is crucial for predicting prognoses and future management strategies.

After experiencing a recurrence, most patients with gastric cancer have a poor prognosis and the majority dies within 3 years (4,5). However, post-recurrence survival (PRS) time is variable among individual patients. Currently, most studies have dealt with prognostic factors for OS or disease-free survival (DFS) in gastric cancer; thus, many clinicopathological parameters including age, sex, histology, number of metastatic and retrieved lymph nodes (LNs), inflammatory markers, or nutritional risks beyond the tumor-node-metastasis (TNM) stage are currently available for predicting survival outcomes (6,7,8,9). However, few studies to date have focused on PRS in patients with recurrent gastric cancer and little is known about predictive factors that affect a patient's prognosis after recurrence.

Positron emission tomography/computed tomography (PET/CT) with F-18 fluorodeoxyglucose (¹⁸F-FDG) has been widely used for staging, evaluating treatment response, and detecting disease recurrence in gastric cancer (10). Notably, ¹⁸F-FDG uptake reflects the biological aggressiveness of gastric cancer; increases have been found to be an independent prognostic marker for patient outcomes in terms of OS or DFS (11). Several studies demonstrated the prognostic value of semi-quantitative ¹⁸F-FDG uptake of the primary tumor or metastatic LN in patients with various stages of gastric cancer (12,13,14,15). However, it remains uncertain whether the glycolytic activity of recurrent tumors could provide prognostic information regarding recurrent gastric cancer. Although diagnostic value of surveillance or restaging using ¹⁸F-FDG PET/CT has been reported by previous studies (16,17,18), there are no reports focused on metabolic activity derived from the ¹⁸F-FDG PET/CT restaging scan as a prognostic marker for PRS in patients with recurrent gastric cancer.

Therefore, we aimed to investigate the prognostic impact of the maximum standardized uptake value (SUV_max) measured by ¹⁸F-FDG PET/CT restaging scans to predict 3-year PRS in patients with recurrent gastric cancer after curative surgical resection.

MATERIALS AND METHODS

This study followed the medical research protocols and ethical guidelines laid down by the World Medical Association's Declaration of Helsinki. The retrospective study protocol was approved by the Institutional Review Board (#2018-06-028), and the need for written informed consent was waived.

Patients

Study participants were selected from 1101 patients with stomach cancer who received potentially curative gastrectomy at our institution between January 2008 and December 2011. The exclusion criteria were non-curative surgery (microscopic or macroscopic residual disease [R1-R2] after resection), the presence of distant metastasis, the reception of any other treatment prior to surgery, a history of previous malignancy, the presence of synchronous malignancy, no recurrence during follow-up, or the absence of ¹⁸F-FDG PET/CT restaging scans. Finally, 47 patients with recurrent gastric cancer who underwent ¹⁸F-FDG PET/CT restaging scans were enrolled in this study.

All patients received total or subtotal gastrectomy along with D2 lymphadenectomy (advanced gastric cancer [AGC]) and D1 + β or D2 lymphadenectomy (early gastric cancer [EGC]). Patients had routinely been followed up every 3 months for the first year after surgery. Subsequently, patients with EGC were followed up every 6 months until 3 years while those with AGC were followed up every 6 months until 5 years. Finally, they were followed up annually using clinical and laboratory examinations with imaging and endoscopic evaluations.

Clinicopathologic and Survival Data

Clinicopathologic data, including sex, age, body weight at surgery and recurrence, percentage of weight loss, surgical and perioperative findings (e.g., type of gastrectomy, pathologic T [pT], pathologic N [pN] and TNM stages, histopathological subtypes, Lauren histotypes, ratio of the number of metastatic LNs to the total number of harvested LNs [LNR]), laboratory values at recurrence, and survival were reviewed and documented. The pT, pN, and TNM stages were classified according to the 8th American Joint Committee on Cancer staging system (19). Neutrophil counts, lymphocyte counts, platelet counts, and hemoglobin levels were obtained at the time of recurrence.

The date of recurrence was defined as follows: the date of imaging examination when imaging findings were used for a definitive diagnosis or the date, when an imaging modality showed abnormal findings for the first time when recurrence was histologically confirmed (20). PRS was defined as the time from the date of recurrence to the date of death; the remaining patients were censored at the last follow-up date that occurred in our institution.

¹⁸F-FDG PET/CT Scan and Image Analysis

All participants performed ¹⁸F-FDG PET/CT scans using two integrated PET/CT scanners (Discovery STE; GE Healthcare, Milwaukee, WI, USA or Biograph mCT; Siemens Healthineers Knoxville, TN, USA). Before ¹⁸F-FDG injection, all patients fasted for at least 6 hours and the blood glucose level of < 150 mg/dL was maintained. Patients were encouraged to rest during the ¹⁸F-FDG uptake period. Images were acquired 60 minutes after 5.5 MBq/kg (Discovery STE) or 4.0 MBq/kg (Biograph mCT) of FDG was administered intravenously. A low-dose CT scan (Discovery STE; peak voltage of 120 kVp and slice thickness of 3.75 mm, Biograph mCT; peak voltage of 120 kVp and slice thickness of 3 mm) was acquired, and PET scan was obtained with an acquisition time of 3 min/bed position with the Discovery STE and 1.5 min/bed position with the Biograph mCT in 3-dimensional mode. Images were reconstructed via ordered-subset expectation maximum iterative reconstruction with attenuation correction.

The images were retrospectively interpreted on an Advantage Workstation 4.3 (GE Healthcare) by two board-certified nuclear medicine physicians. Both readers had knowledge of all available imaging studies; however, they were blinded to the patients' survival data. For the semiquantitative analysis, SUV_max was measured by manually placing circular regions of interest over the visually discernable ¹⁸F-FDG avid metastatic lesions on the attenuation-corrected transaxial ¹⁸F-FDG PET images. To evaluate patients with anastomotic recurrence, abnormally increased uptake at the anastomosis site corresponding to endoscopic and histopathological findings was considered as a recurring malignant lesion. The SUV_max was calculated using the following formula: SUV_max = maximum activity in the region of interest (MBq/g)/ (injected dose [MBq]/body weight [g]).

Statistical Analyses

Numeric data are expressed as medians and interquartile ranges (IQRs), while categorical variables are reported as numbers and percentages. The optimal cutoff values for continuous variables for 3-year PRS predictions were derived from maximally selected chi-square statistics using R package ‘Maxstat’ (21). Recurrence timing was divided into early (≤ 2 years from the surgery date) and late (> 2 years from the surgery date) (22). The Kaplan-Meier method was used to estimate the 3-year PRS rate. All p values < 0.05 were considered statistically significant. For investigating predictive parameters affecting 3-year PRS, multivariate Cox proportional-hazards regression models were performed with the stepwise approach. Variables with p value < 0.05 in the univariate analysis were selected for multivariate analysis; the hazards ratio (HR) and 95% confidence interval (CI) were estimated for each parameter. Statistical analyses were performed using MedCalc for Windows, version 18.10.2 (MedCalc Software, Ostend, Belgium) and R version 3.4.3 software (http://www.r-project.org, R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

Patient Characteristics

In total, 47 patients with recurrent gastric cancer who received curative surgical resection were retrospectively analyzed. Recurrence was confirmed histologically in 33 (70.2%) of the 47 patients, and clinically diagnosed in the remaining 14 patients (29.8%). Overall, 39 of the 47 patients (83.0%) were confirmed dead within 3 years after recurrence during the median follow-up period of 30.3 months (IQR, 18.1–62.6 months), and 3-year PRS rates were 17.0%.

The median PRS time was 10.0 months (IQR, 4.7–19.8 months) in all patients, 50.7 months (IQR, 40.3–71.8 months) in patients who survived over 3 years while 7.7 months (IQR, 4.5–12.4 months) in those who died within 3 years. The median SUV_max obtained from restaging ¹⁸F-FDG PET/CT scans were 4.0 (IQR, 3.3–4.7) in patients surviving over 3 years and 7.3 (IQR, 5.0–9.9) in those who died within 3 years.

The characteristics of the enrolled patients (median age, 59.0 years; IQR, 46.0–64.5 years) are listed in Table 1. Most patients were diagnosed with AGC (89.4%) while only five patients had EGC (10.6%). Pathologic TNM stage III was most frequently observed (63.9%). According to the World Health Organization classification, 31 patients (66.0%) were categorized as having adenocarcinoma (30 tubular and 1 mucinous types) and 16 (34.0%) had a signet ring cell type. In terms of the Lauren histotype, 33 patients (70.2%) were diffuse and 14 (29.8%) were intestinal.

Table 1. Patient Characteristics Depending on 3-Year Post-Recurrence Survival Status.

Variables	Total (n = 47) n (%) or Median (IQR)	Live More than 3 Years after Recurrence (n = 8) n (%) or Median (IQR)	Death within 3 Years after Recurrence (n = 39) n (%) or Median (IQR)	P
Age at diagnosis, years	59.0 (46.0–64.5)	55.5 (47.5–65.5)	59.0 (46.0–64.0)	0.723
Sex				0.146
Male	31 (66.0)	3 (37.5)	28 (71.8)
Female	16 (34.0)	5 (62.5)	11 (28.2)
Type of gastrectomy				0.536
Total	13 (27.7)	1 (12.5)	12 (30.8)
Subtotal	34 (72.3)	7 (87.5)	27 (69.2)
Pathologic T stage*				0.328
T1	5 (10.6)	2 (25.0)	3 (7.7)
T2	6 (12.8)	1 (12.5)	5 (12.8)
T3	7 (14.9)	2 (25.0)	5 (12.8)
T4	29 (61.7)	3 (37.5)	26 (66.7)
Pathologic N stage*				0.203
N0	12 (25.5)	4 (50.0)	8 (20.5)
N1	3 (6.4)	1 (12.5)	2 (5.1)
N2	6 (12.8)	0 (0.0)	6 (15.4)
N3	26 (55.3)	3 (37.5)	23 (59.0)
TNM stage*				0.178
I	5 (10.6)	2 (25.0)	3 (7.7)
II	12 (25.5)	3 (37.5)	9 (23.1)
III	30 (63.9)	3 (37.5)	27 (69.2)
Histopathologic subtype				0.759
Adenocarcinoma	31 (66.0)	6 (75.0)	25 (64.1)
Signet ring cell	16 (34.0)	2 (25.0)	14 (35.9)
Lauren histotype				0.343
Diffuse	33 (70.2)	4 (50.0)	29 (74.4)
Intestinal	14 (29.8)	4 (50.0)	10 (25.6)
LNR	0.11 (0.01–0.33)	0.01 (0.00–0.33)	0.11 (0.05–0.33)	0.311
SUVmax	6.3 (4.2–9.2)	4.0 (3.3–4.7)	7.3 (5.0–9.9)	0.015
Weight loss, %	11.6 (8.6–15.8)	10.3 (5.0–13.9)	11.6 (9.8–17.5)	0.130
Hemoglobin, g/dL	11.7 (11.1–12.6)	11.8 (11.6–12.1)	11.7 (10.9–12.6)	0.671
Neutrophil count, cells/uL	3503 (2601–4894)	2742 (2298–3497)	3934 (2691–5338)	0.125
Lymphocyte count, cells/uL	1317 (964–1739)	1235 (869–1875)	1335 (964–1739)	0.588
Platelet count, x103 cells/uL	228 (193–275)	221 (191–268)	231 (193–278)	0.887
Recurrence timing				0.035
Early, ≤ 2 years	30 (63.9)	2 (25.0)	28 (71.8)
Late, > 2 years	17 (36.1)	6 (75.0)	11 (28.2)
First sites of recurrence				0.192
Locoregional recurrence only	6 (12.8)	2 (25.0)	4 (10.3)
Distant metastasis only	31 (65.9)	6 (75.0)	25 (64.1)
Locoregional and distant failure	10 (21.3)	0 (0.0)	10 (25.6)

^*According to 8th AJCC staging system. AJCC = American Joint Committee on Cancer, IQR = interquartile range, LNR = ratio of number of metastatic lymph nodes to total number of harvested lymph nodes, SUV_max = maximum standardized uptake value, TNM = tumor-node-metastasis

The median time to recurrence (i.e., DFS) was 17.8 months (IQR, 12.4–35.9 months), and a significant portion of patients (63.9%) recurred within 2 years. Six of the 47 patients (12.8%) experienced locoregional recurrence, 31 (65.9%) had distant metastasis, and 10 (21.3%) developed both locoregional and distant failure at the time of recurrence. Most frequent patterns of distant metastasis were hematogenous (60.0%), followed by peritoneal recurrence (18.3%), and then lymphatic metastasis (16.7%). The therapeutic aim of recurrent disease was as follows: 12 patients (25.5%) underwent potentially curative treatment, 27 patients (57.5%) underwent palliative treatment, and the remaining eight patients (17.0%) received only supportive care.

Uni- and Multivariate Analyses

The optimal cutoff values of patients' age, LNR, SUV_max, weight loss percentage, and hemoglobin level as well as for the neutrophil, lymphocyte and platelet counts for a 3-year PRS were 50, 0.063, 5.1, 14.7, 11.4, 2997, 1084, and 154000, respectively. SUV_max performed a Kaplan–Meier analysis with a log-rank test to compare 3-year PRS stratification. Notably, high SUV_max was associated with a significantly lower 3-year PRS rate compared to low SUV_max (3.5% vs. 38.9%, p < 0.001) (Fig. 1).

Fig. 1. Cumulative PRS curves of 47 patients with recurrent gastric cancer stratified by SUV_max.

High SUV_max was associated with significantly lower 3-year PRS rate compared to low SUV_max (3.5% vs. 38.9%, p < 0.001). PRS = post-recurrence survival, SUV_max = maximum standardized uptake value

Values above and below the optimal cutoff for LNR, SUV_max, weight loss, neutrophil count, and recurrence timing were significantly associated with 3-year PRS in the univariate Cox proportional-hazards regression analysis (Table 2). In the multivariate analysis, SUV_max (HR, 2.57; 95% CI, 1.16–5.69; p = 0.012), weight loss (HR, 2.24; 95% CI, 1.11–4.56; p = 0.025), and neutrophil count (HR, 2.68; 95% CI, 1.32–5.43; p = 0.006) were independently prognostic for 3-year PRS. However, the LNR and recurrence timing were no longer statistically significant in the multivariate analysis. A visual presentation of independent prognostic factors for each study participant is displayed in Figure 2.

Table 2. Univariate and Multivariate Analyses for 3-Year Post-Recurrence Survival.

Variables	Univariate Analysis		Multivariate Analysis
	HR (95% CI)	P	HR (95% CI)	P
Age at diagnosis, years
< 50*
≥ 50	1.57 (0.80–3.06)	0.187
Sex
Male*
Female	0.54 (0.27–1.09)	0.086
Type of gastrectomy
Total*
Subtotal	0.69 (0.35–1.36)	0.282
Pathologic T stage†
T1*
T2	1.35 (0.32–5.71)	0.682
T3	0.86 (0.21–3.61)	0.837
T4	1.77 (0.53–5.89)	0.355
Pathologic N stage†
N0*
N1	0.78 (0.17–3.69)	0.754
N2	2.39 (0.79–7.13)	0.119
N3	2.01 (0.89–4.51)	0.091
TNM stage†
I*
II	0.95 (0.26–3.53)	0.942
III	1.79 (0.54–5.97)	0.337
Histopathologic subtype
Adenocarcinoma*
Signet ring cell	1.26 (0.65–2.43)	0.493
Lauren histotype
Diffuse*
Intestinal	0.68 (0.33–1.41)	0.298
LNR
≤ 0.063*
> 0.063	2.22 (1.12–4.43)	0.023
SUVmax
≤ 5.1*
> 5.1	3.30 (1.58–6.87)	0.001	2.57 (1.16–5.69)	0.012
Weight loss, %
≤ 14.7*
> 14.7	2.91 (1.48–5.71)	0.002	2.24 (1.11–4.56)	0.025
Hemoglobin, g/dL
≥ 11.4*
< 11.4	1.88 (0.98–3.58)	0.055
Neutrophil count, cells/uL
≤ 2997*
> 2997	2.89 (1.45–5.79)	0.003	2.68 (1.32–5.43)	0.006
Lymphocyte count, cells/uL
≤ 1084*
> 1084	0.58 (0.30–1.10)	0.095
Platelet count, cells/uL
≥ 154000*
< 154000	2.16 (0.88–5.31)	0.094
Recurrence timing
Early, ≤ 2 years	2.40 (1.18–4.91)	0.016
Late, > 2 years*
First sites of recurrence
Locoregional recurrence only*
Distant metastasis only	1.49 (0.52–4.31)	0.458
Locoregional and distant failure	2.62 (0.81–8.41)	0.107

^*Reference of categorical parameter, ^†According to 8th AJCC staging system. CI = confidence interval, HR = hazard ratio

Fig. 2. Visual summary of independent prognostic factors in series of 47 patients.

Red color indicates patients who died within 3 years after recurrence, and blue color indicates those who survived. Overall survival time after recurrence and numeric data of each parameter are depicted on 4-color scale.

DISCUSSION

In the current study, we assessed the prognostic value of SUV_max on ¹⁸F-FDG PET/CT in patients with recurrent gastric cancer after curative surgical resection. Several studies have reported the prognostic factors for PRS in a variety of malignancies, including stomach, breast, hepatocellular, cervical, and non-small-cell lung cancers (23,24,25,26). However, to the best of our knowledge, no studies have established the prognostic value of ¹⁸F-FDG PET/CT after recurrence in the field of gastric cancer. Our results demonstrated that the SUV_max was an independent survival predictor for 3-year PRS in the multivariate analysis.

Death after recurrence usually occurs rapidly even after achieving curative intent resection. Despite more than 80% of patients dying within 3 years following a recurrence (4,5), there were some long-term survivors. In our study, 17% of the patients lived more than 3 years after relapse and their median PRS was over 50 months. In that regard, predicting patient outcomes after recurrence might be helpful for creating a personalized therapeutic approach or in follow-up planning. Hence, our results suggest a potential value of SUV_max for further prognostication in patients with recurrent gastric cancer. The SUV_max, a simple measurement and the most widely used metabolic parameter obtained by ¹⁸F-FDG PET/CT (27), consistently estimated 3-year PRS outcomes. This may be because increased ¹⁸F-FDG uptake reflects biological aggressiveness not only the initial staging of gastric cancer but also in cases of recurrence (13,28). Thus, ¹⁸F-FDG PET/CT seems to play an additional prognostication role in patients with recurrent disease regarding their survival and in detecting recurrence.

Inflammatory response to cancer contributes to carcinogenesis and tumor progression. Therefore, inflammatory biomarkers have recently been evaluated as valuable prognostic factors in various type of cancers (29); moreover, the neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR) have been widely evaluated for prognostication in gastric cancers (30). Some studies reported NLR to be more predictive for prognoses than PLR (31,32). More recently, Guner et al. (9) demonstrated that simple parameters (e.g., neutrophil count) are better for predicting short- and long-term outcomes in patients with surgically resected gastric cancers compared to complex parameters (e.g., NLR). Our findings show that high neutrophil counts upon recurrence were associated with poor prognoses in patients with relapsed gastric cancer.

The percentage of weight loss, as a simple nutritional parameter, was also an independent survival predictor for PRS in the present study. Weight loss can be an indicator of malnutrition and is closely associated with patients' postoperative quality of life (33). Aoyama et al. (34) found that postoperative weight loss ≥ 15% could lead to worse survival outcomes through a significantly decreased compliance of adjuvant treatments. As appropriate nutritional care could improve survival after recurrence and significantly reduce patients' perioperative morbidity and mortality (35), further studies applying appropriate nutritional assessment tools might help validate whether the patient's nutritional status could affect PRS or not.

Increased glucose consumption and glycolysis are critical hallmarks of gastric cancer; therapeutically targeting tumor cell metabolism processes, such as glucose metabolism, is more convenient approach and is associated with fewer side effects than targeting other biologic systems since cellular metabolic pathways represent the terminus of biologic systems and control the other systems genetically (36,37). Thus, anti-tumor therapies targeting this aspect of cancer are new and appear promising. There have been several studies on therapies reducing glucose consumption of gastric cancer cells (38), suppressing hexokinase II (39,40), or blocking the Warburg effect in combination with other therapies (41) in in vitro and preclinical in vivo settings. As glycolysis-targeted or other novel anti-tumor therapies are discovered, the application of ¹⁸F-FDG PET/CT, a clinical imaging surrogate for enhanced glucose metabolism, could be a considerable strategy to evaluate therapeutic responses and prognostication in recurrent gastric cancer.

The present study had some limitations. The retrospective nature and the relatively small patient dataset from a single institution were the main limitations; these might also have subjected the study to selection bias. Moreover, patients with recurrent cancer who did not undergo ¹⁸F-FDG PET/CT scans were not included. However, our study suggests the potential value of ¹⁸F-FDG uptake for further prognostication in patients with recurrent gastric cancer. In addition, two different PET/CT scanners (Discovery STE and Biograph mCT) were used. The difference in the resolution and administered ¹⁸F-FDG doses could have affected SUV_max values. However, a prior study has validated that the difference in the SUV_max values of the same lesion between two different scanners is < 0.05 (42). Despite these limitations, we suggest that ¹⁸F-FDG uptake at recurrence might be helpful in predicting survival outcomes in patients with recurrent gastric cancer. Further large-scale prospective studies should be conducted using ¹⁸F-FDG PET/CT to assess the prognostic value in recurrent gastric cancer.

In conclusion, the present study revealed that SUV_max could estimate survival outcomes in patients with recurrent gastric cancer after curative resection. Thus, restaging ¹⁸F-FDG PET/CT scans could be used to estimate life expectancy after recurrence and may bolster the role of ¹⁸F-FDG PET/CT in oncology practice.