Low metabolic activity in primary gastric adenocarcinoma is associated with resistance to chemoradiation and the presence of signet ring cells

Abstract

Purposes:

Preoperative chemoradiation is a potential treatment option for localized gastric adenocarcinoma (GAC). Currently, the response to chemoradiation cannot be predicted. We analyzed the pretreatment maximum standardized uptake value (SUV_max) and total lesion glycolysis (TLG) on positron emission tomography/computed tomography as potential predictors of the response to chemoradiation.

Methods:

We analyzed the SUV_max and TLG data from 59 GAC patients who received preoperative chemoradiation. We used logistic regression models to predict a pathologic complete response (pCR) and Kaplan-Meier curves to determine overall survival among patients with high and low SUV_max or TLG.

Results:

Twenty-nine patients (49%) had Siewert type III adenocarcinoma and 30 (51%) had tumors located in the lower stomach. Forty-one patients had poorly differentiated GAC, and 26 had signet ring cells. The median SUV_max was 7.3 (range 0–28.2) and the median TLG was 56.6 (range 0–1881.5). Patients with signet ring cells had a low pCR rate, as well as a low SUV_max and TLG. In the multivariable logistic regression model, high SUV_max was a predictor of pCR (odds ratio = 11.1, 95% confidence interval = 2.12–50.0, p = 0.004). Overall survival was not associated with the SUV_max (log-rank p = 0.69) or TLG (log-rank p = 0.85)

Conclusion:

A high SUV_max was associated with sensitivity to chemoradiation and pCR in GAC, and signet ring cells seemed to confer resistance.

Introduction

Gastric adenocarcinoma (GAC) is estimated to be the fifth most common cancer in the world (951,000 cases) and the third leading cause of cancer-related death worldwide (723,000 deaths) [1]. Perioperative treatment seems useful for localized GAC [2]. Preoperative chemoradiation for GAC is a potential treatment option that is currently being investigated in two prospective randomized trials (TOPGEAR trial, NCT01924819; CRITICS-II trial, NCT02931890) [2–5]. Theoretically, preoperative chemoradiation could increase the R0 resection and/or pathologic complete response (pCR) rates; both outcomes are associated with favorable prognosis [6]. Clinical variables and/or biomarkers associated with the response to therapy are desirable; however, none currently exist.

18-fludeoxyglucose (18F-FDG) positron emission tomography/computed tomography (PET/CT) is occasionally used for GAC staging, especially for detecting metastatic disease [7]. Previous reports have evaluated the prognostic value of the standardized uptake value (SUV) in GAC; however, the findings were inconsistent [8–13]. Total lesion glycolysis (TLG) is gaining more attention because it combines the volumetric and metabolic information obtained from FDG PET/CT. It represents the whole tumor metabolic activity based on the product of the meanSUV and the metabolic tumor volume [14–16]. In contrast, the SUV is often reported as one number corresponding to the maximum glucose consumption (maximum SUV [SUV_max]). TLG has also been evaluated in GAC [17–19]. However, no report has assessed these PET parameters (SUV_max and/or TLG) in conjunction with the response to chemoradiation in the preoperative setting.

In the present study, we analyzed the tumor metabolic activity in 59 patients with GAC who underwent chemoradiation followed by surgery to determine whether these PET parameters could be associated with survival or the histological response.

Methods

Patient cohort

We searched a prospectively maintained database of patients diagnosed with upper gastrointestinal cancer in the Department of Gastrointestinal Medical Oncology at The University of Texas MD Anderson Cancer Center (Houston, Texas, USA) and initially retrieved the records of 550 patients who were diagnosed with GAC (gastric or Siewert type III gastroesophageal junction adenocarcinoma) between March 2002 and April 2015. A total of 132 patients underwent preoperative chemoradiation followed by surgery. Among these patients, 75 underwent PET/CT for initial staging; however, the PET/CT images were unavailable for TLG evaluation in 16 patients. Thus, the records of 59 patients were ultimately included in the study (Figure 1). The institutional review board approved this retrospective analysis.

Figure 1.

The patient selection process for the present study.

PET, positron emission tomography; TLG, total lesion glycolysis.

For each patient, we determined overall survival, which was defined as the time between the date of the diagnosis and the date of death. All patients underwent extensive baseline staging, including computed tomography (CT) and PET/CT studies, esophagogastroduodenoscopy with endoscopic ultrasonography, and blood tests. Staging laparoscopy was performed to assess any peritoneal metastases. Tumor staging was assessed by the American Joint Committee on Cancer Staging Manual (8th edition) [20]. Throughout this article, the term “prognostic marker” is used in the context of the REMARK Guidelines [21]. All patients were evaluated by a multidisciplinary team consisting of medical oncologists, surgical oncologists, radiation oncologists, gastroenterologists, pathologists, radiologists, and various supporting team personnel.

Histological evaluation

The histological features were determined based on the examination of a pretreatment biopsy sample. If any signet cells were present in the specimen, the specimen was classified as signet ring carcinoma. The percentage of signet ring cells in these specimens was not determined.

The extent of residual cancer in the resected specimen was evaluated and assigned to one of three categories: 0% residual cancer (P0); 1–50% residual cancer (P1); or >50% residual cancer (P2) [22].

Treatment strategy

Our treatment approach for localized GAC involves 8 weeks of chemotherapy followed by chemoradiation and then D2 surgery [6, 23]. Induction chemotherapy consists of four doses of 5-fluorouracil and oxaliplatin administered every 2 weeks, and chemoradiotherapy consisting of 45 Gy in 25 fractions with concurrent 5-fluorouracil or capecitabine with or without another cytotoxic agent, such as a platinum compound or taxane (especially when the gastroesophageal junction is involved). Approximately 5 to 7 weeks after the completion of chemoradiation, the patients undergo a preoperative staging workup that includes an endoscopic biopsy and a CT scan with or without a PET scan. Following this, D2 dissection is attempted. All patients in the present study were followed for 3–5 years.

Image acquisition

¹⁸F-FDG PET/CT studies were performed at our institution using an integrated PET/CT system (DST, DRX, or DSTE; GE Healthcare, Milwaukee, WI) before any treatment was administered. CT images were acquired during shallow breathing for attenuation correction (120 kVp, 300 mA, 0.5-second rotation, 3.75-mm slice thickness). A total of 333–629 MBq of intravenous ¹⁸F-FDG was administered to patients after 6 hours of fasting, and PET images were acquired using either the two-dimensional or three-dimensional mode, at 60–90 minutes after the injection of FDG at 3–5 minutes per bed position. Images were reconstructed using ordered subset expectation maximization in two-dimensional mode or iterative reconstruction in three-dimensional mode on standard vendor-provided workstations.

Image analysis

A commercially available software program (MIM Software, Cleveland, OH) was used to delineate the primary tumor as the volume of interest using a semi-automatic gradient-based method. The SUV_max and TLG were extracted from the volume of interest.

Statistical analysis

All statistical analyses were conducted in SAS and Splus. All p values were two-sided and p values of ≤ 0.05 were considered to indicate statistical significance. The chi-squared test was used to assess the association between categorical variables and the Wilcoxon rank-sum test or Kruskal-Wallis test were used to compare the SUV_max and TLG between subgroups of patients defined by patient characteristics. Overall survival was defined as the time interval between the diagnosis and the date of death due to any cause. The probability of overall survival was estimated using the Kaplan-Meier method. The log-rank test was used to compare overall survival between subgroups of patients. Univariate and multivariable logistic regression analyses were performed to identify predictors of a response to chemoradiation, and odds ratios with 95% confidence intervals were calculated. The cutoff values for the SUV_max and TLG that provided optimal differentiation of the response were selected using the recursive partitioning method [24].

Results

Patient characteristics

The characteristics of the 59 patients included in this analysis are listed in Table 1. Twenty-nine patients had Siewert type III gastroesophageal junction adenocarcinoma and 30 had tumors in the distal stomach. The tumors in 38 patients were classified as diffuse type, and 26 patients had signet ring cells in their tumors. Four patients had stage IV disease, identified by positive peritoneal cytologic findings on laparoscopic staging; however, these patients proceeded to local therapy (chemoradiation and surgery) only after their peritoneal cytology became negative on subsequent laparoscopic staging.

Table 1.

Demographic and clinical characteristics of the patients (n = 59)

Clinical feature	No. (%)
Median age (range)	60 years (26–86 years)
Sex
Male	43 (73)
Female	16 (27)
BMI (range)	29.4 (19.1–48.3)
Location of tumor
Siewert type III	29(49)
Distal stomach	30 (51)
Tumor differentiation
Well differentiated	0 (0)
Moderately differentiated	18 (31)
Poorly differentiated	41 (69)
Lauren classification
Intestinal	21 (36)
Diffuse	38(64)
Signet ring cells
Yes	26 (44)
No	33 (56)
Induction chemotherapy
Yes	52(88)
No	7 (12)
Baseline T category
T1	0 (0)
T2	5 (8)
T3	52 (88)
T4	2 (3)
Baseline N category
NO	25(42)
N1	15(25)
N2	11 (19)
N3	8 (14)
Baseline clinical stage
I	4 (7)
IIA	1 (2)
IIB	19 (32)
III	31 (52)
IV	4 (7)
Response
P0	10 (17)
P1	35 (59)
P2	14 (24)
Primary SUVmax (range)	7.3 (0–28.2)
Primary volume (range)	15.2 (0–168.5)
Primary TLG (range)	56.6 (0–1881.5)

Metabolic activity and patient characteristics

For the entire cohort, the median SUV_max was 7.3 (range 0–28.2) and the median TLG was 56.6 (range 0–1881.5). The relationship between metabolic activity and the patient characteristics is shown in Table 2. Both high SUV_max and high TLG were associated with proximal GAC (Siewert type III), the absence of signet ring cells, high clinical T category, node-positive disease, and a high baseline clinical stage (p < 0.05 for all). Although signet ring cells were associated with metabolic activity, tumor differentiation and Lauren classification were not (Figure 2). This could be due to the small number of patients. Eight patients (14%) had no SUV uptake. Among these patients, five had GAC with signet ring cells. pCR (P0) was associated with a high SUV_max (p = 0.009) but not with high TLG (p = 0.21; Figure 2).

Table 2.

Patient characteristics according to metabolic activity (SUV_max and TLG) (n = 59)

Clinical feature	SUVmax			TLG
	Total no.	Median (range)	p value	Median (range)	P value
Age			0.86		0.89
≤median	31	7.94 (0–28.21)		85.41 (0–1881.53)
>median	28	6.47 (0–26.95)		56.22 (0–834.41)
Sex			0.05		0.02
Male	43	8.28 (0–28.21)		106.81 (0–1881.53)
Female	16	5.50 (0–26.95)		17.97 (0–834.41)
Body mass index			1.00		0.39
≤median	29	7.32 (0–28.21)		86.27 (0–1881.53)
>median	30	7.50 (0–23.44)		45.62 (0–834.41)
Location of tumor			0.002		0.01
Siewert type III	29	10.96 (0–26.95)		241.80 (0–1881.53)
Distal stomach	30	5.66 (0–28.21)		27.14 (0–907.91)
Tumor differentiation			0.05		0.37
Well/moderate	18	9.88 (0–26.95)		146.97 (0–759.92)
Poor	41	6.02 (0–28.21)		54.93 (0–1881.53)
Lauren classification			0.06		0.36
Intestinal	21	9.47 (0–26.95)		135.15 (0–759.92)
Diffuse	38	5.99 (0–28.21)		55.41 (0–1881.53)
Signet ring cells			0.004		0.02
Yes	26	5.80 (0–23.31)		41.04 (0–834.41)
No	33	9.47 (0–28.21)		241.80 (0–1881.53)
Baseline T category			0.02		0.004
T1/2	5	0 (0–7.38)		0(0–16.77)
T3/T4	54	7.83 (0–28.21)		89.79 (0–1881.53)
Baseline N category			0.01		0.02
N0	25	5.56 (0–26.95)		34.64 (0–759.92)
N1/2/3	34	8.67 (0–28.21)		150.23 (0–1881.53)
Baseline clinical stage			0.01		0.004
I/IIA/IIB	24	5.35 (0–26.95)		25.40 (0–759.92)
III/IVA	35	9.05 (0–28.21)		158.79 (0–1881.53)
Response			0.009		0.21
P0	10	14.62 (0–28.21)		260.06 (0–907.91)
P1	35	6.53 (0–24.18)		54.93 (0–1881.53)
P2	14	5.78 (0–17.2)		45.62 (0–834.41)

SUV_max, maximum standardized uptake value; TLG, total lesion glycolysis; BMI, body mass index.

Figure 2.

The histological features, response to chemoradiation, and metabolic activity. SUV_max, maximum standardized uptake value (upper panel); TLG, total lesion glycolysis (lower panel). P0, 0% residual cancer; P1, 1% to 50% residual cancer; P2, >50% residual cancer.

Metabolic activity and response

To identify the factors predicting pCR, we performed a logistic regression analysis with the patients categorized according to the SUV_max (low, <13.4; high, ≥13.4) and TLG (low, <156; high, ≥156). In the univariate logistic regression model, the absence of signet ring cells (odds ratio [OR] = 9.37, 95% confidence interval [CI] = 1.10–79.71), high SUV_max (OR = 14.28, 95% CI = 2.94–50.0), and high TLG (OR = 5.88, 95% CI = 1.32–25.0) were associated with pCR (Table 3). In the multivariable logistic regression analysis, which only included the clinical features that were identified as significant in the univariate analysis (absence of signet ring cells, high SUV_max, and high TLG), high SUV_max was the only independent predictor of pCR (OR = 11.1, 95% CI = 2.12–50.0, p = 0.004; Table 3). The absence of signet ring cells and high TLG were not associated with pCR in the multivariable model (Table 3).

Table 3.

Univariate and multivariable logistic regression models for metabolic activity and a pathologic complete response (0% residual cancer)

Clinical feature		Univariable		Multivariable with SUV		Multivariable with TLG
		Odds ratio (95% CI)	p value	Odds ratio (95% CI)	p value	Odds ratio (95% CI)	p value
Agea		0.97 (0.92 – 1.03)	0.33
Sex	Male	1 (reference)	0.21
	Female	0.25 (0.02 – 2.17)
Location of tumor	Siewert type III	1 (reference)	0.45
	Stomach	0.59 (0.15 – 2.35)
Tumor differentiation Well/moderate		1 (reference)	0.48
	poorly	0.60 (0.15 – 2.46)
Lauren classification	Intestinal	1 (reference)	0.30
	Diffuse	0.49 (0.12 – 1.92)
Signet ring cells	YES	1 (reference)	0.04	1 (reference)	0.11	1 (reference)	0.13
	No	9.37 (1.10 –79.71)		6.44 (0.68 – 61.46)		5.62 (0.59 – 43.73)
Baseline T category	T1/2	1 (reference)	0.85
	T3/T4	0.80 (0.08 – 8.02)
Baseline N category	N0	1 (reference)	0.39
	N1/2/3	1.90 (0.44 – 8.23)
Baseline clinical stage	I/IIA/IIB	1 (reference)	0.45
	III/IVA	1.75 (0.40 – 7.58)
Primary SUVmax	<13.4	1 (reference)	0.001	1 (reference)	0.004
	≥13.4	14.28 (2.94 – 50.0)		11.1(2.12 – 50.0)
Primary TLG	<156	1 (reference)	0.02			1 (reference)	0.14
	≥156	5.88 (1.32 – 25.0)				3.33 (0.67 – 16.7)

SUV_max, maximum standardized uptake value; TLG, total lesion glycolysis; CI, confidence interval.

^aContinuous variable.

Metabolic activity and survival

Twenty-three of the 59 patients (39%) died. There was no significant difference in overall survival between the low and high SUV_max groups (p=0.69); or between the low and high TLG groups (p=0.85) (Figure 3).

Figure 3.

The Kaplan-Meier analysis of overall survival with patients categorized according to the maximum standardized uptake value (SUV_max; low, <13.4; high, ≥13.4; left panel) and total lesion glycolysis (TLG; low, <156; high, ≥156; right panel).

Discussion

Our results demonstrate that high SUV_max was a significant independent predictor of pCR after preoperative chemoradiation, while TLG was not. This could mean that the high level of metabolic activity reflected by the SUV_max is likely associated with responsiveness to chemoradiation rather than TLG, which reflects the entire tumor area. However, other explanations are possible and will need to be explored. GACs with signet ring cells often have low metabolic activity due to the low expression of the GLUT1 receptor and the use of glutamine rather than glucose as an energy source [25]. The low metabolic activity in GAC might represent the presence of a high volume of signet ring cells, which would result in chemoradiation resistance [26].

No previous report has evaluated the relationship between metabolic activity and the response to preoperative chemoradiation in GAC. However, the relationship between metabolic activity and the response to chemotherapy has previously been evaluated in GAC. Park et al. showed that GACs with a low SUV_max were sensitive to chemotherapy in the metastatic setting [11]. Several reports noted that the reduction in SUV_max after chemotherapy is a reflection of the response to chemotherapy in GAC [17, 27–29]. In the case of chemoradiation, PET evaluations after treatment are influenced by inflammatory changes; therefore, it is difficult to evaluate metabolic change [30, 31]. Our group previously showed that the change in SUV_max was prognostic in esophageal cancer after induction chemotherapy but not after chemoradiation [32]. This study is the first to show that the pretreatment SUV_max was independently associated with pCR in GAC patients who received trimodality therapy.

Previous studies have shown that GAC with signet ring cells tends to have low metabolic activity, which is consistent with the findings in the present study [8, 18, 25, 33]. A previous small cohort study showed that ¹⁸FDG-PET was less useful for predicting recurrence in signet ring cell carcinoma than in non-signet ring cell carcinoma [10]. These results suggest that the histological features should be considered when ¹⁸FDG-PET is used to predict the prognosis or the response to preoperative treatment.

The mechanism by which signet ring cells cause therapeutic resistance has not been fully elucidated. Diffuse-type GAC contains rich stromal cells and creates tumor microenvironments (TME) that might be associated with therapeutic resistance [34]. For instance, cross-talk between the tumor cells and cancer-associated fibroblasts has been found to confer therapeutic resistance due to epithelial–mesenchymal transition [35]. Further studies are warranted.

The present study was associated with some limitations. First, this was a retrospective study with a relatively small population. Second, the results of the PET studies, which are not routinely performed in patients with GAC, were missing in some patients, who were therefore excluded from our analysis.

Nevertheless, our results suggest that a high SUV_max is associated with pCR in GAC patients who undergo chemoradiation followed by surgery. Our results also suggest that signet ring cell carcinoma has low metabolic activity and is associated with resistance to therapy.