Cyclo-oxygenase-2 expression is associated with mean standardised uptake value on 18F-Fluorodeoxyglucose positron emission tomography in oesophageal adenocarcinoma

Abstract

Objectives:

This pilot study investigated the association of four PET image features and cyclo-oxygenase-2 (COX-2) expression in patients with oesophageal adenocarcinoma. The prognostic significance of these biomarkers was also assessed.

Methods:

50 consecutive patients [median age = 68 (range 47 – 84), males = 45) with oesophageal adenocarcinoma had PET/CT staging between January 2011 and July 2015. The maximum and mean standardised uptake values (SUV_max and SUV_mean), metabolic tumour volume (MTV) and tumour lesion glycolysis (TLG) were calculated from the primary tumour. Their association with COX-2 status was assessed using Mann-Whitney U tests. Kaplan-Meier and Cox regression analysis tested their prognostic significance. A p-value < 0.05 was considered statistically significant.

Results:

32 tumours (64.0%) were COX-2 positive. There was a significant association between SUV_mean and COX-2 status (p = 0.019). TLG (hazard ratio (HR) 1.001, 95 % confidence intervals (CI) 1.000 – 1.002, p = 0.018) was significantly associated with overall survival on multivariable analysis.

Conclusions:

This study investigated the association between PET image features and COX-2 expression in oesophageal adenocarcinoma. The preliminary results signal that a combination of TLG (calculated as product of MTV and SUV_mean) and COX-2 status may be a strong and clinically important prognostic biomarker. Our research group are planning a prospective, multi-centre study to validate these findings.

Advances in knowledge:

Mean standardised uptake value (SUV_mean) on PET imaging is associated with COX-2 expression in oesophageal adenocarcinoma.

Introduction

The prognosis of oesophageal cancer is poor and adenocarcinoma is the most common histological cell type in developed countries.¹ The development of oesophageal adenocarcinoma has been linked with prolonged mucosal inflammation causing progression through the metaplastic to dysplastic to adenocarcinoma sequence.²

Cyclo-oxygenase-2 (COX-2) is involved in promoting angiogenesis and is an enzyme that is activated in response to extra-cellular stimuli such as pro-inflammatory cytokines.³ It is over expressed in epithelial solid cancers associated with inflammation. Non-steroidal anti inflammatory drugs (NSAIDS) may reduce the risk of cancer in the gastrointestinal (GI) tract and aspirin has become the focus of preventative clinical trials.⁴ Importantly, COX-2 has been shown to have prognostic significance.⁵

Positron emission tomography (PET) combined with CT using 18F fluorodeoxyglucose (FDG) is now routinely used in the oesophageal cancer staging pathway. The focus of many PET research studies in oesophageal cancer is the identification of prognostic imaging biomarkers. Our research group have found that PET image features, maximum and mean standardised uptake value (SUV_max and SUV_mean), metabolic tumour volume (MTV) and tumour lesion glycolysis (TLG) have prognostic significance,⁶ with similar results confirmed in other studies.^7,8

Prognostic biomarker studies that correlate immunohistochemical tumour marker expression with imaging biomarkers are currently lacking in oesophageal cancer research, however an association between the SUV_max and COX-2 has been found in lung adenocarcinoma.⁹ The discovery of combined prognostic radiological and pathological biomarkers may improve patient risk stratification and treatment decision making. Improved patient selection for more personalised treatment regimens may ultimately improve the currently low survival rates.

Therefore, the primary aim of this pilot study was to obtain preliminary data associating PET image features (SUV_max, SUV_mean, MTV and TLG) and COX-2 expression in oesophageal adenocarcinoma. The secondary aim was to assess the prognostic significance of these biomarkers.

methods and materials

Patient selection

This retrospective pilot study considered all consecutive patients with biopsy-proven oesophageal and gastro-oesophageal junction adenocarcinoma in a single tertiary centre who had staging PET/CT between January 2011 and July 2015 (n = 71). Patients were radiologically staged using the TNM seventh edition.¹⁰ Given that consecutive patients were studied, the patients included in the study received a variety of surgical, oncological and palliative treatments. (Table 1) Ethical approval was granted to quantify COX-2 expression from archived tissue and correlate with routinely performed PET/CT (REC 14/WA/1208). The requirement for informed consent was waived.

Table 1.

Baseline Characteristics of the Patient Cohort

Characteristic	Frequency (%)
M: F Ratio	45 (90.0): 5 (10.0)
Tumour Location
Oesophagus  GOJ	23 (46.0) 27 (54.0)
Adenocarcinoma Differentiation
Well  Moderate  Poor  Gx	2 (4.0) 16 (32.0) 19 (38.0) 13 (26.0)
Radiological T-stage
T1  T2  T3  T4a  T4b  Tx	2 (4.0) 3 (6.0) 34 (68.0) 7 (14.0) 2 (4.0) 2 (4.0)
Radiological N-stage
N0  N1  N2  N3	13 (26.0) 14 (28.0) 15 (30.0) 8 (16.0)
Radiological M-stage
M0  M1	40 (80.0) 10 (20.0)
Stage Groups
Stage I  Stage II  Stage III  Stage IV	3 (6.0) 10 (20.0) 27 (54.0) 10 (20.0)
Treatment
Palliative  NACT  dCRT  Surgery alone  NACRT	23 (46.0) 14 (28.0) 6 (12.0) 5 (10.0) 2 (4.0)

GOJ, Gastro-oesophageal junction; Gx/Tx, unable to be assessed; NACRT, neo-adjuvant chemoradiotherapy; NACT, neo-adjuvant chemotherapy; dCRT, definitive chemoradiotherapy.

Patients with insufficient archived tissue from the biopsy sample (n = 17) and cases in which the primary tumour was non-avid on PET/CT (n = 4) were excluded. 50 patients were included in the study.

PET image features

The standard PET/CT protocol in our centre has previously been published in detail⁶ and is included in Supplementary Material 1. This identical protocol was used for each patient in this study (Figure 1). Data preparation was performed by a single researcher, a radiology resident with 5 years’ experience of PET research who was blinded to COX-2 status. Primary oesophageal tumour delineation was performed using the Automatic decision Tree-based Learning Algorithm for Advanced image Segmentation in positron emission tomography (ATLAAS) tool,¹¹ which eliminates inter-observer variation in tumour outline. SUV_max, SUV_mean, MTV and TLG were automatically calculated from the primary tumour only using software developed and validated by our research group. Further details are included in Supplementary Material 1. The SUV_max is the value extracted from the pixel with the highest uptake value in the ATLAAS-defined region of interest. The SUV_mean is calculated as the average uptake value across the MTV. TLG is calculated as the product of SUV_mean and MTV.

Figure 1.

A selected fused PET/CT image showing a FDG-avid distal oesophageal adenocarcinoma.

COX-2 preparation

COX-2 analysis was performed from archived diagnostic biopsy tissue because this biomarker is not routinely tested during the staging pathway. The tissue was obtained prior to PET/CT and treatment initiation. COX-2 staining was performed on the Leica Bond III automated immunostaining platform and detection carried out using the Leica Bond Polymer Refine DAB system. Primary antibody COX-2 mouse monoclonal (Clone CX-294, (Dako, Ely, UK)) was applied.

COX-2 quantification

COX-2 expression was quantified by a Consultant Upper GI Histopathologist and classified as a categorical variable based on the intensity of staining using the following grading: 0, 1+, 2 + and 3+¹² (Figure 2). A grading of 0/1 + was considered negative and 2+/3 + considered positive. A pre-constructed control tissue microarray (TMA) was added to the slides as a control with standardised 3+, 2+, 1 + and 0 scoring tumour for COX-2. Internal negative controls within non-tumour components of the biopsies were also tested.

Figure 2.

A high magnification image of an adenocarcinoma biopsy sample showing high COX-2 expression (brown cells).

Survival data

The secondary outcome of the study was overall survival, defined in months from the date of diagnosis. No patient was lost to follow-up, with each patient receiving clinical follow-up 3-monthly in the first year, then 6-monthly thereafter for the next 4 years, or until death.

Statistical analysis

Categorical variables were described as frequency (percent) and continuous variables as median (range). Mann-Whitney U tests were used to assess for significant differences between PET image features and COX-2. Chi-square tests assessed differences between TNM stage and COX-2 status. Kaplan-Meier analysis with log-rank test evaluated the association of TNM stage and COX-2 status with overall survival. Univariable and multivariable Cox regression models tested the association of continuous PET image features with overall survival. The multivariable model included TNM stage, SUV_max, SUV_mean, MTV, TLG and COX-2 status. A p-value < 0.05 was considered statistically significant. Statistical analysis was performed using SPSS v23.0 (IBM, Chicago, IL).

Results

Patient cohort

Baseline characteristics of the patient cohort are detailed in Table 1. The median age of patients was 68.0 years (range 47–84). 32 tumours were classified as positive COX-2 status (2 += 18, 3 += 14) and 18 tumours were classified as negative COX-2 status (0 = 4, 1 += 14). Most patients (46%) were treated palliatively.

PET image features

Four PET image features were calculated from the ATLAAS-defined primary tumour. The median SUV_max was 13.49 (min 3.56, max 42.05, interquartile range (IQR) 11.85). The median SUV_mean was 7.16 (min 2.07, max 34.97, IQR 5.50). The median MTV was 19.09 mL (min 0.66, max 94.76, IQR 20.13). The median TLG was 125.33 (min 11.25, max 1809.08, IQR 234.69). No significant difference in MTV or TLG was found between patients treated with radical intent compared with those treated palliatively (p = 0.376 and p = 0.224, respectively).

Association of TNM stage and PET image features with COX-2 status

SUV_mean (U = 172.0, mean rank 31.9 vs 21.9, p = 0.019) showed a significant association with COX-2 status. (Figure 3) A higher grade of COX-2 expression correlated with a lower SUV_mean value. SUV_max, MTV and TLG did not reach statistical significance with COX-2 status (p = 0.079, p = 0.486 and p = 0.203, respectively).

Figure 3.

Box-plot representations of COX-2 status with SUV_mean (p = 0.019).

Prognostic significance of TNM stage, PET Image Features and COX-2 Status

MTV (hazard ratio (HR) 1.024, 95% confidence intervals (CI) 1.007–1.041, p = 0.005) and TLG (HR 1.001, 95% CI 1.000–1.002, p = 0.011) were both significantly associated with overall survival on univariable analysis. T-stage (X² 24.998, df 5, p < 0.001) and N-stage (X² 12.201, df 3, p = 0.007) were significantly associated with overall survival. As there were only 10 cases with M1 disease, M-stage had borderline prognostic significance (X²3.151, df 1, p = 0.076). SUV_max (p = 0.803) and SUV_mean (p = 0.838) were not associated with overall survival. COX-2 status was also not associated with overall survival (X² 0.010, df 1, 0.921) (Figure 4). After 36 months, the negative and positive status groups appeared to separate, but this was not statistically significant, possibly a reflection on the small sample size. On multivariable analysis, T-stage (p = 0.008) and TLG (p = 0.018) were independently and significantly associated with overall survival (Table 2).

Figure 4.

Cumulative survival of negative and positive COX-2 status groups (X² 0.010, df 1, p = 0.921).

Table 2.

Results of Multivariable Analysis

				95% CI
Variable	p-value	Hazard ratio	df	Lower	Upper
T-Stage	0.008		5
T2	0.487	2.272	1	0.224	23.039
T3	0.878	0.853	1	0.112	6.506
T4a	0.705	1.548	1	0.162	14.821
T4b	0.216	4.861	1	0.398	59.406
Tx	0.019	25.017	1	1.694	369.534
TLG	0.018	1.001	1	1.000	1.002

Tx unable to be assessed, df degrees of freedom, CI confidence intervals, TLG Tumour Lesion Glycolysis

Discussion

This study presents preliminary data regarding COX-2 expression and PET image features in oesophageal adenocarcinoma. SUV_mean was associated with COX-2 status, with positive COX-2 expression linked to lower FDG-uptake values. In addition, MTV on univariable analysis, and T-stage and TLG on multivariable analyses were significantly associated with overall survival in this patient cohort.

These findings suggest that metabolic activity on FDG-PET may be influenced by COX-2 expression. Variation in FDG-uptake is associated with underlying pathophysiological features such as angiogenesis, vascularity, perfusion, hypoxia and necrosis.¹³ Furthermore, COX-2 has a role in angiogenesis and subsequently tumour vasculature. This pilot study demonstrated an unexpected inverse association between FDG-uptake and COX-2 status. Previous research has suggested that COX-2 may be over expressed in more aggressive tumours which tend to be associated with higher FDG-uptake.¹⁴ However, this inverse association has been found in other tumour sites including breast.¹⁵

The cause of this inverse association is not fully understood. CT perfusion studies in oesophageal and colorectal cancer have investigated tumour vasculature.¹⁶ This dynamic technique quantifies tumour perfusion and can predict response to neo-adjuvant treatment. In oesophageal cancer, increased blood flow and reduced mean transit time are associated with tumour regression grade. To date, no in vivo studies have investigated COX-2 expression and tumour perfusion in oesophageal cancer, but increased COX-2 may promote angiogenesis, resulting in varied tumour blood flow affecting tumour growth and hence reduced metabolic activity. Alternatively, variation in angiogenesis may contribute to intra tumoural necrosis and reduced FDG-uptake. These hypotheses may explain the inverse relationship identified here but the relationship between COX-2 expression and FDG-uptake remains controversial.¹⁴

COX-2 expression is prognostically significant. In one systematic review and meta-analysis in oesophageal adenocarcinoma, COX-2 had the largest survival effect (HR 2.47, 95% CI 1.15–3.79) out of 9 tumour markers including human epidermal receptor growth factor 2 (HER-2) and Ki-67.⁵ In this current study, MTV on univariable and TLG on multivariable analyses were significantly associated with overall survival. These results support other published research.⁷ As expected, disease stage was also significantly associated with overall survival. Tumours with larger MTV and higher FDG-uptake are likely to indicate more advanced tumours and therefore be associated with poorer outcomes. Although COX-2 was not prognostically significant in this patient cohort, preliminary data was obtained from 50 consecutive patients only. This pilot study is under-powered to detect this difference and a larger cohort may have demonstrated statistical significance between COX-2 status groups. In addition, treatment regimens did not influence survival differences between these groups, with relatively equal proportions of patients receiving radical or palliative treatments.

Given the proven link of increased cancer risk with inflammation, the association with COX-2 status and FDG-uptake on PET could signal the potential for clinically important combined biomarkers. Aspirin has become the focus of preventative clinical trials given the link between non-steroidal anti inflammatory drug (NSAID) use and inflammation in cancer.⁴ COX-2 is thought to be over expressed in areas of inflammation such as segments of Barrett’s oesophagus.² Our research group is planning a prospective, multi-centre study to further investigate and validate the association of COX-2 status and PET image features in oesophageal adenocarcinoma. Such research should follow the key recommendations outlined in the imaging biomarker roadmap for cancer studies, which describes the necessary stages in the translational pathway of potential image biomarkers from discovery to validation and adoption into clinical practice.¹⁷

This pilot study had several strengths. All patients underwent PET/CT using an identical protocol and scanner. This sample of patients were treated as part of a large, experienced Regional Upper GI cancer MDT, serving a population of over 1.4 million. All diagnostic biopsies and tumour markers were evaluated by a Consultant GI Histopathologist. No patients were lost to follow-up.

Limitations

This retrospective, single centre pilot study with a small patient cohort has a number of limitations which should be addressed in future studies. As for similar studies involving biopsy data, it is possible that results are affected by differences in tumour marker expression between the sample and the whole imaged tumour volume. Multiple populations of clonal cells are known to exist within tumours, therefore independent biopsies sampling different regions of malignant tissue may result in varying levels of tumour marker expression. In addition, tumour marker analysis was conducted from archived tissue. Older tissue may degrade affecting its quality and quantity¹⁸ but cases in which there was insufficient tissue were excluded from this study. The oldest biopsy in our cohort was 4 years old.

Conclusion

This pilot study associated PET image features and COX-2 expression in oesophageal adenocarcinoma. This study provides preliminary evidence that SUV_mean is associated with COX-2 status and further evidence that MTV and TLG are prognostically significant. These findings suggest that further research is warranted and will inform a prospective, multi-centre study investigating PET image features and COX-2 expression.