Initial Standardized Uptake Value (iSUV) of Positron Emission Tomography Influences Prognosis of Patients with Localized Gastric Adenocarcinoma Treated Preoperatively

Abstract

Background

In patients with localized gastric adenocarcinoma (LGAC) who receive preoperative therapy, tools to predict response or prognosticate outcome before therapy are lacking. We used initial standardized uptake value (iSUV) of positron emission tomography (PET) to evaluate its association with overall survival (OS).

Methods

We identified 60 patients with confirmed LGAC who were treated with preoperative chemoradiation and had a baseline PET in addition to other routine staging. Fisher’s exact test and Wilcoxon’s rank sum test were used to determine the association between iSUV and other variables and log rank test and Cox proportional hazards model were used for survival analysis.

Results

Median iSUV was 6 (range, 0–28). The presence of signet ring cells in pretreatment biopsies correlated highly with low iSUV≤6 (p=0.0017). Patients with high iSUV>6 had a longer OS compared to those with low iSUV≤6 (p=0.0344). iSUV was not an independent prognosticator (p=0.12), however the risk of death was reduced for patients with high iSUV>6 (HR=0.26).

Conclusion

Our novel findings show that among LGAC patients treated with preoperative chemoradiation and surgery, those with high iSUV have longer OS than patients with low iSUV. iSUV appears to have a prognostic role in patients with LGAC when treated with preoperative chemoradiation.

INTRODUCTION

Gastric adenocarcinoma (GAC), a common malignancy with a poor prognosis, is listed as the third leading cause of cancer-related mortality worldwide. [1] In the United States 24,590 new cases are estimated to be diagnosed and 10,720 cases to succumb to GAC in 2015. [2] Surgical resection plays a major part in cure of patients with localized GAC (LGAC), however 5-year survival rates after surgery alone are 20–50% and adjunctive therapy must be offered. [3] Multimodality treatment options for GAC include postoperative chemoradiation, perioperative chemotherapy and postoperative chemotherapy. In addition, multiple phase II trials have evaluated the role of preoperative chemoradiation for GAC. [4] With preoperative chemoradiation, the rates of pathologic complete response (pathCR; no residual tumor cells in the resected surgical specimen) range from 20–25% and 70–78% of patients are able to undergo an R0 resection. [5–7]

Patients with LGAC, when treated with preoperative therapy, tend to have heterogeneous and unpredictable outcomes. Heterogeneity in patients’ outcomes is frequently noted even if similarly staged patients receive the same therapy. This heterogeneity is most likely related to the differing molecular biology of individual GAC, responsible for different levels of sensitivity to chemotherapy and radiotherapy and its metastatic potential. [8]

In addition to improving outcomes by identifying novel agents, there is a need to customize preoperative therapy of LGAC patients based on GAC biology. Evidence suggests that imaging modalities such as positron emission tomography (PET), using [¹⁸F]-fluorodeoxyglucose (FDG) as a tracer and integrating with computed tomography (CT) (PET/CT) for anatomic localization, seems to be an effective non-invasive tool to predict biological aggressiveness and prognosis of patients with LGAC. However, its role in defining biology is debatable as other variables can play roles as well.

Prior studies are conflicting. Initial standardized uptake value (iSUV) has been correlated with overall survival (OS) after surgery. Others either found correlation between low iSUV and OS or lack of it. [9–13] Two studies have addressed iSUV and histopathologic subtypes of GAC. [12, 13] To our knowledge, ours is the first study examining iSUV as prognosticator in LGAC patients who received preoperative chemoradiation and we provide compelling reason for our results.

MATERIALS AND METHODS

Patient Selection

We identified patients from our prospectively assembled and maintained GAC database in the Department of Gastrointestinal Medical Oncology at The University of Texas MD Anderson Cancer Center (UTMDACC). The Institutional Review Board of UTMDACC approved the analysis. We studied 60 consecutive patients with confirmed LGAC who received preoperative therapy and surgery between 2002 and 2013. All had a baseline PET/CT. All patients were staged with upper endoscopy with endoscopic ultrasound as well. All patients had a negative peritoneal staging by laparoscopy or laparotomy. Clinical staging was based on the American Joint Committee on Cancer (AJCC) Classification, 6^th edition. [14]

Trimodality therapy

All patients had concurrent chemotherapy with radiation. All patients received a fluoropyrimidine (intravenously or oral) with or without either a taxane or a platinum compound during radiotherapy. The total radiation dose delivered was either 45 Gy in 25 fractions or 50.4 Gy in 28 fractions, at 1.8 Gy per fraction delivered once daily, 5 days per week prescribed to cover 95% of a clinical target volume encompassing the primary tumor, mucosal margin and regional lymphatic regions. Prior to chemoradiation, 53 patients (88%) received up to 8 weeks of induction chemotherapy.

Approximately 5 to 7 weeks after the completion of chemoradiation, all patients underwent preoperative upper endoscopy with biopsies and a PET/CT or CT scan of chest and abdomen. All patients proceeded to surgery. Surgical procedures included extensive lymph node dissection and specimens were staged by AJCC, 7^th edition. [15]

Follow-up and survival

Patients were monitored periodically until 5 years after surgery or until death. Follow-up survival data were obtained from the hospital records, patient follow-up data, death records (social security database) and the UTMDACC tumor registry. The estimated median follow-up time was 2.64 years (95% CI: 2.34 – 4.36 years).

FDG-PET/CT scans

PET/CT scans were performed on a dedicated PET/CT system (Discovery ST-8; GE Medical Systems, Milwaukee, WI). Details of PET scanning and SUV calculations have been described in detail from our group elsewhere. [16–19] At our institution, we follow the NCI guidelines for image preparation, acquisition and analysis. [20–22]

Statistical Methods

Fisher’s exact test and Wilcoxon’s rank sum test were used to determine the association between iSUV and other variables. The log rank test and Cox proportional hazards model were used for survival analysis. For the purpose of exploratory, a multivariate Cox model was fitted to evaluate whether iSUV was significantly associated with OS after adjusting for the effects of EGOC PS, baseline lymph node status, histologic grade and presence of SRC. Variables with a p value < 0.05 were considered statistically significant. SAS software 9.3 (SAS institute, NC, Cary) and Splus software 8.2 (TIBCO, CA, Palo Alto) were used for the analyses.

RESULTS

Patient and tumor characteristics

The baseline and post-treatment patient characteristics are shown in Table 1. Patients were primarily men (63%), Caucasians (67%), with median age of 60 years (range, 26–82 years). Twenty-nine had type 3 gastroesophageal junction adenocarcinoma according to Siewert classification and 31 had GAC elsewhere.

Table 1.

Summary of baseline characteristics

Covariate	Levels	Frequency (percent)
Race	Caucasian	40(66.7%)
	Other	20(33.3%)
Gender	Female	22(36.7%)
	Male	38(63.3%)
ECOG PS	0	16(26.7%)
	1	42(70%)
	2	2(3.3%)
Siewert classification	Type 3 GEJ	29(48.3%)
	Gastric	31(51.7%)
Baseline EUS T	T2	5(8.3%)
	T3/T4	55(91.7%)
Baseline EUS N	N0	27(45%)
	N1/N2	33(55%)
Baseline M	M0	60(100%)
Baseline Clinical Stage	I/II	26(43.3%)
	III/IV	34(56.7%)
Presence of Signet ring cells (SRC)	Yes	31(51.6%)
	No	29(48.3%)
Histologic Grade	G2-Moderately differentiated	17(28.3%)
	G3-Poorly differentiated	43(71.7%)
iSUV	≤6	30(50%)
	>6	30(50%)
Tumor Size	N/A-Not available	18(.%)
	<5	18(42.9%)
	≥5	24(57.1%)

Initial SUV and patient characteristics

For the analysis, iSUV was dichotomized by 6, which was the median iSUV. Fisher’s exact test was used to evaluate the association between iSUV and other variables. Presence of SRC correlated highly with low iSUV (p=0.0017). Fisher’s exact test suggested that patients with type 3 LGAC (69%) were more likely to have a higher iSUV (>6) compared to those located elsewhere in the stomach (32%) (p=0.0092). Also, larger tumors (≥5cm) were more likely to have high iSUV (p=0.0043). There was a non-significant trend of association between low iSUV and high histologic grade (G3) (p=0.084). iSUV could not predict pathologic complete response (p=0.1042).

Table 2 shows the results of Wilcoxon’s rank sum test to compare continuous variables with iSUV>6 and iSUV≤ 6. LGACs were larger in patients with an iSUV>6 compared to those with an iSUV≤ 6 (p=0.012).

Table 2.

Wilcoxon’s rank sum test to compare continuous variables between patient groups with iSUV≤ 6 and iSUV>6

Covariate	iSUV	n	Mean +/− std, median (range)	Wilcoxon rank sum test p value
Age at Presentation at MDACC	H (>6)	30	57.9 +/− 12, 59 (26 – 82)	0.4987
	L(≤6)	30	60 +/− 10.9, 60.5 (36 – 78)	.
Tumor Size (cm)	H (>6)	25	6.5 +/− 3.3, 6 (1.2 – 18)	0.0120
	L(≤6)	17	4.4 +/− 2.6, 4 (1.4 – 12)	.
Baseline BMI	H (>6)	11	28.6 +/− 13.4, 25.7 (0 – 48.3)	0.5732
	L(≤6)	15	26.5 +/− 6.5, 29.7 (17.3 – 34.5)	.
SUV uptake of primary	H (>6)	30	13 +/− 5.6, 11.8 (6.4 – 28.2)	<0.0001
	L(≤6)	30	2.8 +/− 2.2, 3.4 (0 – 6)	.

Initial SUV and survival

Log rank test was used to compare OS among subgroups of patients. Patients with high iSUV>6 had longer OS compared to those with low iSUV≤6 (p=0.0344). The only other factor strongly associated with shorter OS in the log rank test was poor performance status (ECOG PS 2; p=0.0002).

In the exploratory multivariate analysis, iSUV was not independently prognostic (p=0.12) after adjusting for the effects of EGOC PS, baseline lymph node status, histologic grade and presence of SRC (data not shown). However, the risk of death was reduced for patients with high iSUV>6 (Hazard Ratio for death=0.26). Patients with low iSUV≤6, G3 histology, and SRC had a tendency for shorter OS.

DISCUSSION

SUV of the PET reflects the biology of cancer and possibly metabolic rate. Prior studies reported poor OS and high recurrence rates in patients with high iSUV but these patients had primary surgery. [11, 12] Stahl et al. [9] reported that OS was not different in patients with high or low iSUV after preoperative chemotherapy. Similar results in a later study by Vallböhmer et al. [23] showed that iSUV did not correlate with OS after preoperative chemotherapy. Table 3 shows all the studies evaluating the potential of iSUV for the assessment of prognosis in the LGAC setting. In our study of patients that received preoperative chemoradiation, those with high iSUV tended to do better.

Table 3.

Studies assessing the influence of iSUV of PET/CT in the localized gastric cancer setting

Ref.	No of patients	Type of treatment	Findings
Stahl et al. (9)	40	Preoperative Chemotherapy	No significant correlation with prognosis
Vallböhmer et al. (23)	42	Preoperative Chemotherapy	No significant correlation with prognosis
Mochiki et al. (11)	85	Primary Surgery	High iSUV → Worse OS
Pak et al. (13)	41 (with SRC histology)	Primary Surgery	High iSUV → Worse aggressive clinical course, although iSUV not independently prognostic
Lee JW et al. (12)	271 (with subgroup analysis according to histopathologic subtypes)	Primary Surgery	High iSUV → Predictive factor for tumor recurrence for patients with TAC or PAC, not for SRC and MAC

iSUV: Initial standardized uptake value; PET/CT: Positron emission tomography with computed tomography; OS: Overall survival; SRC: Signet ring cell; TAC: Tubular adenocarcinoma; PAC: Poorly differentiated adenocarcinoma; MAC: Mucinous adenocarcinoma.

Our study sheds light on histologic subtypes and the SRC showing significant association with iSUV. Prior reports using the Lauren classification have shown that FDG-PET/CT had significantly lower sensitivity in detecting diffuse type, SRC, and mucinous GACs than intestinal type adenocarcinomas. [9, 24–27] Low iSUV as a poor prognosticator may be due to lack of avidity in diffuse type of GAC [9, 28] and low expression of glucose transporter 1 (GLUT1). [9, 25, 29]

GLUT1, the human erythrocyte glucose transporter, is a member of the facultative glucose transporters family, which currently is comprised of 14 members. Previous studies demonstrated the role of overexpression of facultative glucose transporter genes in human cancer. [30] GLUT1, as a glucose metabolism-related protein, is responsible for FDG SUV uptake in various tumors. Overexpression of GLUT1 has been described in tumors of the esophagus, [31] colon, [32] pancreas, [33] lung, [34] brain, [35] ovary [36] and breast. [37]

Rapid cancer growth induces a hypoxic environment in tumors. Hypoxia-inducible factor 1 (HIF1) acts as a sensor for hypoxic stress upregulating angiogenic factors and activating transcription of genes encoding glucose transporters, such as GLUT1. Thus, overexpression of GLUT1 may play an important role for tumor survival by promoting an adequate energy supply through the glycolytic pathway, supporting their high metabolic rate in a not so ideal hypoxic environment. [38]

In gastric cancer, the frequency and level of GLUT1 expression do not appear to be very high; approximately only 30% of cases examined in the study of Kawamura et al [29] were positive. These differences in GLUT1 expression may be attributed to histological differences. The SRC and mucinous GACs, which are characterized by low cellularity and/or high mucinous content, present a very low FDG uptake due to low GLUT1 expression. [29, 39, 40] One possible reason why high FDG SUV uptake by GAC does not reflect tumor aggressiveness like in other neoplasms, as expected, came from a more recent study that implied glutamine metabolism may be upregulated in GAC. [41] Thus, GAC cells could eliminate the necessity for glucose transport as they use glutamine as energy source in a hypoxic tumor microenvironment. This malignant transformation has been reported in other cancers [42] as well and interestingly an ongoing glutamine-based PET phase I study is currently recruiting participants for patients with history of histologically-confirmed solid malignancies and/or lymphomas (NCT01697930).

The positive association between high SUV and lower histologic grade found in our study is not necessarily unexpected. Contrary to other tumors, G2 GACs accumulated FDG more intensely than G3 GACs. Therefore, histologic grade may not represent biological aggressiveness in this tumor. A plausible reason to explain high SUV of G2 tumors in our study is the absence of SRC and mucinous component in these low-grade tumors (0 out of 17 patients had SRC in their pretreatment biopsies), a finding shown also in the study of Stahl et al. [9]

Essentially, in our study, low iSUV indicates SRC or G3 histology and therefore poor outcome in LGAC. On the other hand, high iSUV is associated with no SRC, lower (G2) grade and more sensitivity to chemoradiation, therefore predicting good outcome and prolonged OS. Finally, our data suggest that in our population of LGAC patients who received preoperative chemoradiation, high iSUV is associated with a prolonged OS. In contrast, when single modality such as surgery or surgery with preoperative chemotherapy is used, high iSUV leads to poor patient outcomes and is not a good prognosticator of OS, as shown in previous studies (Table 5), being aware of the difficulty to compare such data between different studies. However, an explanation is that tumors with high iSUV (no SRC and low-grade) are more sensitive to chemoradiation. High iSUV may indicate a group of patients presenting with LGAC with favourable prognosis when treated with preoperative chemoradiation.

Our study being retrospective has some shortcomings: 1) it is a single high-volume center experience, 2) the total denominator is relatively small and because the number of events was also small, the analysis was not powered enough to detect differences in survival and 3) clear and reproducible cut-off values are still lacking and prospective multicenter studies need to be carried out with centralized SUV assessment. But our data have strengths that can contribute to the management of patients with LGAC as they involve a uniformly treated population.

In conclusion, our novel findings show that among LGAC patients treated with preoperative chemoradiation and surgery, those with no SRC and/or G2 histology have high iSUV>6 and longer OS whereas patients with SRC and/or G3 histology have low iSUV≤6 and shorter OS. Thus, iSUV appears to have a prognostic role in patients with LGAC when treated with preoperative chemoradiation.