Positron emission tomography-CT (PET-CT) is one of the initial mediastinal staging modality for non-small cell lung cancer; however, the clinical utility in carcinoid tumours is uncertain. We sought to determine the test performance of PET-CT for mediastinal lymph node staging of pulmonary carcinoid tumours. We collated data from seven institutions, performing a retrospective search on pathological databases for a consecutive series of patients who underwent thoracic surgery (with lymph nodal dissection) for carcinoid tumours with preoperative PET-CT staging. PET-CT results were compared with the reference standard of pathologic results obtained from lymph node dissection and test performance reported using sensitivity and specificity. From November 1999 to January 2013, 247 patients from seven institutions underwent surgery for carcinoid tumours with a corresponding preoperative PET-CT scan. The mean age of the patients was 61 (SD 15, range 73) and 84 were male patients (34%). The pathologic subtype was typical carcinoid in 217 patients (88%) and atypical carcinoid in 30 patients (12%). Results from lymph node dissection were obtained in 207 patients. The calculated sensitivity and specificity of PET-CT to identify mediastinal lymph node disease was 33% (95% CI 4% to 78%) and 94% (95% CI 89% to 97%), respectively. Our results indicate that PET-CT has a poor sensitivity but good specificity to detect the presence of mediastinal lymph node metastases in pulmonary carcinoid tumours. Mediastinal lymph node metastases cannot be ruled out with negative PET-CT uptake, and if the absence of mediastinal lymph node disease is a prerequisite for directing management, tissue sampling should be undertaken.
- Imaging/CT MRI etc
- Thoracic Surgery
- Non-Small Cell Lung Cancer
Statistics from Altmetric.com
Carcinoid tumours are a rare group of malignancies arising from neuroendocrine cells1 in the gastrointestinal tract and bronchopulmonary system. Within the lung, they account for 1%–2% of all malignancies1 and are classified into typical (well differentiated) and atypical (less well differentiated) subtypes.1 Typical carcinoids form approximately two-thirds of cases and metastasise in approximately 12% with an overall survival rate >90%,1 but atypical carcinoids are comparatively rarer and more aggressive accounting for one-third of cases with >50% mediastinal lymph node metastases and 5-year survival rate of 40%–75%.1
As rare tumour subtypes, it is difficult to conduct large scale studies and hence evidence to inform on diagnosis, staging and management is limited. Many centres include patients with carcinoid tumours in management pathways for non-small cell lung cancer (NSCLC) where 18fluoro-dexoy-glucose positron emission tomography-CT (18FDG PET-CT) is widely used to stage the mediastinum.2 Current British Thoracic Society Guidelines recommends radical treatment without further mediastinal lymph node sampling if there is no significant uptake in normal sized mediastinal lymph nodes on 18FDG PET-CT.3
Little is known about the clinical utility of 18FDG PET-CT for mediastinal or distant disease staging in carcinoid tumours as their metabolic activity is often considered low or variable and it is argued whether 18FDG PET-CT serves the same purpose in selecting patients for radical management in carcinoid tumours.
We sought to determine the test performance of 18FDG PET-CT for mediastinal lymph node staging of carcinoid tumours by collating data from member institutions of the UK Thoracic Surgery Research Collaborative.
Data from seven institutions (Royal Brompton & Harefield NHS Foundation Trust, Liverpool Heart and Chest Hospital NHS Foundation Trust, Heart of England NHS Foundation Trust, Derriford Hospital, Bristol Royal Infirmary, Golden Jubilee National Hospital and Castle Hill Hospital) were collated on a consecutive series of patients who underwent thoracic surgery for carcinoid tumours from November 1999 to January 2013. Pathologic and radiological reporting was performed by UK accredited pathologists and radiologists. Preoperative 18FDG PET-CT staging reports were obtained from patient records and the site(s) of lymph node uptake documented. Technical conduct, extent of surgery and operative lymph node dissection were performed according to each surgeon's individual practice. The reference standard in this study was postoperative pathologic reporting of lymph node involvement on samples harvested. It was not possible to intentionally blind the pathologists to results of 18FDG PET-CT as this was a retrospective study.
As clinical decisions are made on the presence or absence of mediastinal lymph node metastases only, the unit of measurement in this study was the binary outcome of positive or negative for mediastinal lymph node disease on 18FDG PET-CT and the corresponding operative pathology reports. Lymph nodes were defined as positive as stated on pathology or PET-CT reports, respectively. No uptake or ‘negative’ annotation on the report was defined as negative. Data compilation was undertaken separately at each institution, anonymised and collated into a central registry.
Patients were excluded from analysis if preoperative 18FDG PET-CT was not performed (or the results unavailable), if the 18FDG PET-CT was performed ≥4 months before surgery, if the postoperative pathological report was not available or if no lymph node dissection was performed.
Continuous data were presented as mean with SD or median with IQR as appropriate to the data distribution. Categorical and count data were presented as frequency and percentage (%). Test performance was assessed as sensitivity and specificity and reported with 95% confidence values. Statistical analyses were undertaken using Stata V.10 (College Station, Texas, USA).
From November 1999 to January 2013, a total of 247 patients from seven institutions underwent surgery for a carcinoid tumour with a corresponding preoperative 18FDG PET-CT scan. We excluded one patient with a ‘false’ positive 18FDG PET-CT result due to TB. The mean age of the patients was 61 (SD 15, range 73) of which 84 were male patients (34%). The pathologic subtype was typical carcinoid in 217 (88%) and atypical carcinoid in 30 (12%). The mean standardized uptake value in the primary tumour was 4.6 (SD 3.8) for typical carcinoids and 6.1 (SD 4.4) for atypical carcinoids. There was a mean of 38 (SD 21) days between 18FDG PET-CT scan and surgery. Surgery was predominantly via lobectomy and in two cases no resection was performed due to patient choice (table 1).
Results from lymph node dissection were obtained in 207 patients. A median of four stations sampled (IQR 2–7). Positive mediastinal lymph uptake was reported on 18FDG PET-CT in 15 patients of whom two were reported to be involved on subsequent pathology (atypical carcinoid in both cases). Of the 207 patients who underwent lymph node dissection, six patients were reported to have mediastinal lymph node metastases (two typical, four atypical carcinoids).
For our primary outcome, the calculated sensitivity and specificity of 18FDG PET-CT to identify mediastinal (N2) lymph node disease were 33% (95% CI 4% to 78%) and 94% (89% to 97%), respectively.
In our secondary analysis for the test performance of 18FDG PET-CT for hilar (N1) disease, 18FDG PET-CT reported uptake at hilar lymph nodes in 21 patients, of whom five were positive on subsequent pathology (four typical, one atypical carcinoids). Pathologic results from hilar lymph node dissection carried out in 207 patients at the time of surgery found 32 patients with hilar lymph node positive disease (22 typical, 10 atypical carcinoids). The sensitivity and specificity of 18FDG PET-CT at identifying hilar lymph node disease were similar at 16% (95% CI 5% to 33%) and 91% (86% to 95%), respectively.
Our results represent the largest cohort to date suggesting that 18FDG PET-CT has poor sensitivity (to rule out) but good specificity (to rule in) for the presence of mediastinal lymph node metastases in patients with typical and atypical carcinoid tumours.
The test performance of sensitivity of 18FDG PET-CT for mediastinal lymph node metastases is considerably poorer than for NSCLC (33% vs 84% NSCLC3) but the specificity is comparable (94% vs 89% NSCLC3). With such a large disparity between the accepted sensitivity of 18FDG PET-CT for mediastinal lymph node staging in NSCLC nodal staging and pulmonary carcinoid tumours, this indicates that further sampling of mediastinal nodes is required in the presence of a negative 18FDG PET-CT result if subsequent management is dependent on the absence of mediastinal lymph node disease.
It is unlikely that a large sample study of the test performance of 18FDG PET-CT is feasible. The results from our study suggested that the test performance for two secondary sites (hilar and mediastinal lymph nodes) was similar and we hypothesise that if the results were to be extrapolated to distant disease, the test performance would be similar.
Due to the multicentre, retrospective nature of our study, it is a limitation that there was no standardised method of mediastinal lymph node management across the participating hospitals. Several of the hospitals practice systematic lymph node dissection, dissecting all stations on ipsilateral side, while others practiced more limited lymph node sampling. Dissections where only one station was removed are obviously of limited benefit when describing the utility of PET-CT to detect the global involvement of lymph nodes. On average, there were a median of four stations sampled (IQR 2–7). We have also not collected any follow-up data in this study and therefore cannot comment on locoregional disease after surgery. This may be an area for future work, allowing us to define our cases which were microscopically negative as true negatives.
Octreotide (or analogue) scintigraphy is increasingly adopted for gastrointestinal neuroendocrine tumours. A recent study comparing the test performance of octreotide scintigraphy with 18FDG-PET-CT in a cohort of 21 patients with pulmonary carcinoid reported cumulative sensitivities and specificities for the primary tumour, lymph nodes and distant metastases of 76% and 97% (octreotide scintigraphy) versus 85% and 89% for 18FDG PET-CT, respectively,4 and concluded that octreotide scintigraphy may not significantly improve the identification mediastinal staging above that of 18FDG PET-CT. As there is uncertainty over the empiric use of either technique for lymph node staging we argue against the routine use of either or both techniques without prior decision on the appropriate management of patients with mediastinal and/or distant disease. The European Neuroendocrine Society Guidelines that are currently being prepared are anticipated to recommend radical management for typical carcinoid tumours even in the presence of mediastinal or distant disease as long as each site is treated with a radical intent and if so we would need to reconsider the value or need for preoperative metabolic imaging as a pure staging investigation.
Our results indicate that 18FDG PET-CT has a poor sensitivity but good specificity to detect the presence of mediastinal lymph node metastases when used for staging pulmonary carcinoid tumours. Mediastinal lymph node metastases cannot be ruled out with negative 18FDG PET-CT uptake and if the absence of mediastinal lymph node disease is a prerequisite for directing management, tissue sampling should be undertaken.
This paper was presented at the 2013 Annual meeting of the American Society of Clinical Oncology (Chicago, USA) and the15th World Conference of Lung Cancer (Sydney Australia).
Contributors EL and HP undertook the study conception and design. All coauthors contributed to collection of data and reporting of the work. EL and HP drafted the article and all authors contributed to the final manuscript.
Funding This project was in part supported by the National Institute of Health Research Respiratory Disease Biomedical Research Unit at the Royal Brompton and Harefield NHS Foundation Trust and Imperial College London.
Competing interests None.
Provenance and peer review Not commissioned; internally peer reviewed.