Accuracy of pre‐operative 18‐fluoride fluorodeoxyglucose positron emission tomography (FDG‐PET) in predicting lymph node involvement in colon cancer

Accurate staging of colon cancer is imperative in directing treatment and prognostication. Existing literature on pre‐operative accuracy of FDG‐PET/CT in detecting lymph node disease often combines colon and rectal cancer, examines rectal cancers alone, and rarely assesses colon cancer in isolation. Our aim was to assess pre‐operative utility of FDG‐PET/CT in detecting lymph node disease in colon cancer.


Introduction
Colorectal cancer was the third most diagnosed cancer in Australia in 2018 and was the second most common cause of cancer death in 2020. 1 Australia has one of the highest incidences of colorectal cancer in the world and globally the disease burden is predicted to not only increase but impact younger patients. 2In cases without nodal involvement, the five-year survival is 90%, but this drops to 71% if there is locoregional spread and decreases to 13% in cases of distant metastases. 3Thus, accurate detection and preoperative staging is critical in overall survival.Colonoscopy is a key component in detecting and diagnosing colonic cancers, 4 however, it does not provide information regarding depth of tumour invasion and lymph node involvement.This requires staging with imaging including contrast enhanced computed tomography (CT) of the chest, abdomen and pelvis, plus dedicated focused imaging such as contrast enhanced liver MRI in the setting of suspicion for metastatic liver disease. 5Carcinoembryonic antigen (CEA) is also a useful quantitative serum marker which in its elevated levels correlate with reduced overall survival especially after resection. 68-Fluoride fluorodeoxyglucose positron emission tomography with CT (FDG-PET/CT) is used to detect increased glucose metabolism or uptake that occurs in colon cancer and uses a radiolabelled glucose analogue to detect tumour foci too small to be confidently diagnosed by CT or other structural imaging alone.By using fused CT this allows anatomical localisation of the areas of increased glucose cellular metabolism.However, physiological gut uptake and false-positive uptake in inflammatory disease precludes method of imaging in primary diagnosis or screening. 7The uptake of the radio-labelled glucose by active cancer is measured as standard uptake value (SUV).Lower tumour uptake of FDG measured as maximum SUV, has been associated with improved survival. 8There have been multiple studies performed demonstrating the efficacy of FDG-PET/CT against CT alone.Overall, there appears to be higher specificity but comparable or lower sensitivity at detecting nodal disease in colorectal cancers at 84% and 44% respectively. 9,10Accordingly, FDG-PET/CT is often used to assess indeterminate distant lesions, and can play a role in the postresection setting when assessing for recurrence of disease, for example in rectal cancer attempting to differentiate between pelvic recurrence and postoperative fibrosis.There is a reported 95% sensitivity and 97% specificity in this setting. 7n Australia, FDG-PET/CT is only government funded following initial therapy, for the evaluation of suspected residual, metastatic or recurrent colorectal carcinoma in patients considered suitable for active adjuvant therapy. 11The primary staging of colonic cancer is important as the presence of distant disease affects the mode and timing of treatments.Contrary to rectal cancer, colon cancer does not currently have an accepted neoadjuvant regime and thus is reliant on surgical management for confirmation of staging and treatment.Our aim was to assess the pre-operative utility of assessing nodal disease with FDG-PET/CT in colon cancer.

Methods
A retrospective cohort analysis was performed on a single Australian tertiary institution database.Data were extracted for all primary colonic tumours who had pre-operative FDG-PET/CT and underwent primary surgical resection between January 2017 and April 2022 in the same institution.Cases were reviewed to include only treatment naïve colon cancer patients.All locally recurrent colon cancers, those undergoing neoadjuvant therapy, and all rectal cancers were excluded.Metachronous tumours were included as long as they had not had treatment within 12 months, with the new primary being considered 'treatment naïve'.All tumours were adenocarcinoma variants.Colonic resections for alternative malignancies were excluded.Patient demographics, operation reports, formal histopathology, medical imaging and preoperative CEA were reviewed on the patient's electronic medical records (eMR) and the data entered into SPSS.
Operations were performed by or under the supervision of four specialist colorectal surgeons each with a minimum 5 years specialist experience.The standard of care in our colorectal surgical unit is to perform a high ligation of the vessel to maximize lymphovascular yield.Final lymph node yield was recorded and operation reports were reviewed to confirm if a high ligation was performed.In particular cases where an inadequate lymph node yield (<12 on histopathology) or false positive result was recorded, the histopathology was reviewed in an attempt to correlate and determine adequacy of surgery and therefore accuracy of FDG-PET/CT findings; if determined inadequate they were excluded.
The primary outcome was sensitivity and specificity of positive nodal disease on pre-operative staging with FDG-PET/CT compared with final surgical histopathology staging.The FDG-PET/CT were reviewed by the institution's Nuclear Medicine group by at least one senior Nuclear Medicine specialist with greater than 5 years' experience.All reporting physicians have completed both local and overseas training.To standardize the avidity of lymph nodes, the SUV cut off to be considered positive was set at >2.5 based on recent literature. 5 A single outside FDG-PET/CT was included and re-reviewed by our local group for consistency of reporting for preoperative planning.
Secondary outcomes were patient and tumour factors predictive of FDG-PET/CT positive disease including pre-operative CEA, mismatch repair (MMR) status, duration from FDG-PET/CT to surgery, and Tumour (T)-stage.The data were analysed using SPSS statistics for Mac (Version 26; SPSS, Chicago, IL, USA).Two by two tables were used to calculate sensitivity, specificity, negative and positive predictive values as well as accuracy.Mann-Whitney U and Chi square tests were used to analyse secondary outcomes.A P-value <0.05 was considered statistically significant.The project protocol was submitted to and reviewed by the South Eastern Sydney Local Health District (SESLHD) Human Research Ethics Committee and was deemed low-negligible risk and subsequently approved to proceed.

Results
A total of 339 patients were identified as potential primary colonic tumours, with 34 patients having had pre-operative FDG-PET/CT.Thirty three of these were performed within our institution.The majority (25/34; 74%) were performed with a documented indication for further staging: 20 of these for investigation of indeterminate lesions on other imaging, although only two of these were for intraabdominal nodal disease.Other indications included surveillance of an unrelated malignancy (2/34; 6%), and concern for locally advanced disease (2/34; 6%).Despite extensive review of the records in three (9%) cases no specific concern or indication other than 'staging' was identified.Table 1 shows a breakdown of the indications.
Of the patients, 20 were male (59%) and the mean age was 65 years.Mean lymph node harvest was 19 nodes (range 8-40), and the mean SUV reported on FDG-PET/CT for the primary cancer was 16.5.The mean SUV of true positive lymph node cases was 9.4, and those that were classified as false positives 5.9.There was no statistically significant difference between these groups (P = 0.44).All patients had a primary adenocarcinoma, with the majority of patients (25/34; 74%) having moderately differentiated subtypes.The median duration between FDG-PET/CT and operative date was 17 days.Patient characteristics are summarized in Table 2.
Pre-operative FDG-PET/CT reported positive lymph node involvement in 12 patients (12/34; 35%).Of these patients nine were true positives on histopathological analysis, whilst three were shown to be false positives.On final histopathology 17 patients had node positive disease.Eight patients who had no evidence of lymph node disease on FDG-PET/CT had positive histopathology (Table 3).Of the two patients whose recorded indication for FDG-PET/CT was other 'imaging concerning for local nodal disease', one was a true positive and one was a true negative.FDG-PET/CT detection of lymph node involvement had a sensitivity of 53%, specificity of 82%, positive predictive value of 75%, negative predictive value 64% and accuracy of 68%.
Further assessment of the false positive cases was performed, with review of the operation reports, clinical notes, imaging and histopathology to assess for a failure to achieve adequate surgical clearance.All three cases were reported to have high ligations, and had lymph node yields of 18, 28 and 36 nodes.One case was likely an omental deposit mistakenly labelled as a lymph node based on the pathology findings.The other two cases had reported areas of mesenteric uptake with SUV of 8.1 and 3.8 respectively but no corresponding disease identified.
Additional assessment was also undertaken of the four cases with a lymph node yield less than 12.They had yields of 8, 10, 11 and 11 nodes.On final histopathology the specimen with the lowest yield of 8 was a true positive.This specimen had been reported as being reviewed twice but despite further sectioning no additional nodes were identified.The other cases were a false negative and two true negatives.
The median pre-operative CEA of those with positive nodes on FDG-PET/CT was 9 μg/L.Mann-Whitney U and Chi square tests showed no significant difference in median pre-operative CEA, MMR status, T-stage or duration between FDG-PET/CT and operative date for those with FDG-PET/CT positive nodal disease (P = 0.287, P = 0.162, P = 0.098, and P = 0.053 respectively).The primary tumour SUV, tumour differentiation or tumour location had no apparent impact on the sensitivity or specificity of FDG-PET/CT in determining lymph node involvement.

Discussion
This study displays an underwhelming sensitivity (53%) and accuracy (68%) of FDG-PET/CT for nodal disease in colon cancer suggesting a need to change the paradigm of its use in staging.The moderate specificity (82%) and positive predictive values (75%) add some utility for those with a positive result, but from the indications identified, particularly the three cases where no indeterminate lesions were being investigated, it suggests a need to re-think the role of FDG-PET/CT as a staging scan in treatment naïve colon cancer.
A key issue in the staging of colon and rectal cancer is the prediction of lymph node involvement.The sensitivity of CT for the detection of lymph node involvement has been reported to range between 29% and 90% for colorectal cancer, compared to a sensitivity of 52% and specificity of 85% associated with FDG-PET/ CT. 9,10 A 2016 meta-analysis involving 13 studies reported  sensitivity and specificity for CT in detecting isolated colon cancer invasion beyond the intestinal wall to be 90% and 69% respectively, and sensitivity and specificity for nodal involvement to be 71% and 67% respectively. 12Another meta-analysis in 2020 concluded that FDG-PET/CT had low sensitivity but high specificity in colorectal cancer. 13This meta-analysis described heterogeneity between the studies, and unfortunately no separation was made with regards to the site of the primary lesion.In reporting sensitivity, specificity and accuracy for isolated colon cancer this study adds to the literature, as studies have infrequently separated results by tumour site or colon versus rectal primary.There are variations in the literature regarding recommendations for the role of preoperative FDG-PET/CT in the staging of nodal disease for isolated colon cancer, with rectal cancer and combined cohorts of 'colorectal' patients being described.A Korean study published in 2014 by Lee et al. involving 266 colon cancer patients found that preoperative use of FDG-PET/CT changed the management plan in 6.5% of patients with clinical stage III colon cancer and 12.7% of patients with clinical stage IV colon cancer. 14With the addition of FDG-PET/CT, 6% of patients were upstaged and 19.9% were downstaged.It showed a high sensitivity but poor specificity for FDG-PET/CT in lymph node detection.Lee et al. noted a change in management was most frequently following the detection of hepatic metastases amenable to surgical resection. 14Furthermore, there is limited literature on the use of FDG-PET/CT for pre-operative nodal staging of colon cancer in an Australian setting, where a universal healthcare system mandates local funding limitations according to Medicare indication.In our institutions practice, it was noted that an ad-hoc and sometimes non-evidence-based approach to pre-operative FDG-PET/CT with a paucity of evidence as to the role of FDG-PET/CT other than examining indeterminate findings or suspicion of metastatic disease.It is useful in characterizing hepatic and pulmonary lesions in patients that have known avid disease, but again there is limited evidence with regards to accuracy of determining nodal disease. 10,14he poor sensitivity and accuracy for nodal disease suggests that for patients without pre-operative suspicion of metastatic or nodal disease there is limited utility in routine pre-operative FDG-PET/ CT use for colon cancer nodal staging.We hypothesised that this may be due to inadequate volume of nodal disease to have sufficient metabolic activity to be detected on FDG-PET/CT compared to other distant metastatic disease.Future larger studies which include volumetric assessment of those with positive nodes should be considered.
A limitation of the study included its retrospective nature.A prospective study had previously been explored at our institution, however, cost and the ethics of exposing patients to a potentially unnecessary and unfunded additional investigation with an imaging modality that uses ionizing radiation had been considered prohibitive.There was no control over the time between imaging and obtaining histopathology, however, with a median duration of 17 days this is unlikely to have affected outcomes.An accepted range of up to 4 months was determined to balance study inclusion size against the possibility of bias.A longer duration potentially increases the false negative rate, which in turn would reduce the calculated sensitivity.
The retrospective nature and inability to control surgical technique limited the ability to control the adequacy of lymph node yield, and the potential for bias.The possibility of a missed positive node cannot be excluded, however, the high lymph node yields reported in the false positive cases make this less likely.False negatives where lymph node yields are low remain a difficulty, however, repeat sectioning of specimens and the finding of true positives and involved nodes despite low yields provide some reassurance.A prospective study could consider standardized criteria to overcome this before patients were enrolled.
Additionally, the study was limited by modest patient numbers (approximately 10% of colonic resections performed during the same period) having had a pre-operative FDG-PET/CT due to it not being routinely funded.The database had limited additional demographic data, and tumour biology was not readily available beyond that described.For example, whilst MMR screening is routine in our institution, BRAF testing only occurs if MMR deficiency is detected.The study size was also felt to have inadequate size for this to lead to reliable conclusions regarding the impact of tumour factors.Furthermore, another limitation is that as funding for a pre-operative FDG-PET/CT is limited to those with concern for metastatic disease.This may have biased towards a higher pre-operative probability of metastatic or nodal disease being present.As outlined in Table 1 most FDG/PET-CT requests were based on indeterminate alternative imaging findings.Additionally, imaging is often obtained for staging in inpatients where the Medicare criteria may not be met.As the cost is absorbed by the hospital, this may have led to bias towards those with an acute issue leading to inpatient admission and concurrent staging investigations.With small patient numbers unfortunately a review of concurrent acute issues did not demonstrate any specific associations.

Conclusion
FDG-PET/CT has moderate specificity but poor sensitivity in the detection of lymph node involvement in colon cancer.Its utility should likely remain isolated to investigating pre-operative equivocal metastatic lesions, or investigation of suspected residual, metastatic or recurrent disease.Given the underwhelming sensitivity demonstrated, this study does not support the routine use of FDG-PET/CT for staging of lymph node involvement in colon cancer.With the paucity of existing evidence and inherent limitations of the retrospective nature of this study, further large prospective studies would be useful to examine predictive factors for positive FDG-PET/CT and define the optimal patient populations for its utility pre-operatively.

Table 1
Indication for FDG/PET-CT

Table 2
Characteristics of 34 patients identified who underwent FDG-PET/CT prior to primary surgical resection of their colon cancer

Table 3
FDG PET/CT results compared to final histopathology