Is there a role for traditional nuclear medicine imaging in the management of pulmonary carcinoid tumours?†

Abstract

OBJECTIVES: The clinical utility of fluorodeoxyglucose-positron emission tomography (FDG-PET) and somatostatin receptor scintigraphy (SRS) in pulmonary carcinoids staging is unclear. This study aims to determine the role of FDG-PET and SRS in detecting hilar-mediastinal lymph node metastasis from these tumours.

METHODS: We retrospectively collected the data of 380 patients who underwent lung resection for primary pulmonary carcinoid in seven centres between 2000 and 2015. Patients without nodal sampling (n = 78) were excluded. In 302 patients [35% men, median age 58 (interquartile range 47–68) years] the results of preoperative computed tomography (CT) scan, FDG-PET and SRS were analysed and compared to the pathological findings after resection to determine the respective utility of these two nuclear tests.

RESULTS: The sensitivity, specificity and negative predictive value in detecting N1 and N2 disease were respectively 33% and 46%, 93% and 90%, 88% and 95% for computed-tomography-scan, 38% and 60%, 93% and 95%, 88% and 95% for FDG-PET, 22% and 33%, 95% and 98%, 84% and 87% for SRS. The diagnostic accuracy for N1 and N2 disease of CT scan was not significantly different from that of FDG-PET (P = 1.0 and P = 0.37 for N1 and N2 disease respectively) and of SRS (P = 0.47 and P = 0.35 for N1 and N2 disease respectively). The sensitivity and specificity of these imaging tests were also similar when analysed by typical vs atypical histology.

CONCLUSIONS: CT scan, FDG-PET and SRS showed similar performance in terms of nodal staging for pulmonary carcinoid. These findings suggest that additional nuclear imaging beyond CT scan is not required as long as a lymphadenectomy or nodal sampling is completed at resection.

Pulmonary carcinoid tumours, Computed tomography scan, Fluorodeoxyglucose positron emission tomography , Somatostatin receptor scintigraphy, Mediastinal staging

INTRODUCTION

Pulmonary carcinoids are well-differentiated neuroendocrine tumours arising from the neuroendocrine cells of the pulmonary systems. They are rare, account for 1–2% of all lung malignancies and are classified as typical carcinoid (TC) or atypical carcinoid (AC) [1]. TC, which represent 65–90% of all lung carcinoids, are indolent neoplasms with a low rate of lymph node and distant metastasis at presentation (5–15% and 3% respectively) and a 5-year overall survival rate >90%. Whereas AC, which represent 10–35% of all lung carcinoids, have a more aggressive behaviour (40–50% and 20% have respectively lymph nodes and distant metastasis at presentation) with 5-year overall survival rate of 40–50% [2]. The first key element in planning the best strategy of care for these two entities is to ensure accurate preoperative diagnosis and staging.

Computed tomography (CT) is currently the main staging modality. However, other functional imaging methods such as fluorodeoxyglucose positron emission tomography (FDG-PET) and somatostatin receptor scintigraphy (SRS) have often been used to stage these tumours [1]. The utility of FDG-PET and SRS has come into question because of the generally low growth rate of carcinoids and the possible lack of somatostatin receptors. These features may lead to inaccurate staging and suboptimal surgical treatment. The aim of this study is to compare the performance of CT, FDG-PET and SRS in detecting hilar and mediastinal lymph node metastasis, in order to assess the role of the two nuclear imaging methods in the lymph nodal staging of these tumours in a multi-institutional cohort undergoing surgical treatment.

MATERIALS AND METHODS

We performed a retrospective, multi-institutional review of consecutive patients who underwent lung resection for histologically confirmed pulmonary carcinoid tumour from 2000 to 2015. Participating institutions in order of the number of cases contributed included the Swedish Cancer Institute in Seattle WA; UC Davis Medical Center in Sacramento, CA; Catholic University ‘Sacred Heart’, Rome, Italy; University of Insubria, Ospedale di Circolo, Varese, Italy; University of Washington Medical Center, Seattle, WA; Providence Regional Medical Center, Everett, WA; Virginia Mason Hospital and Medical Center, Seattle, WA. We excluded patients who did not undergo lymph node sampling or lymphadenectomy. This study was approved by the institutional review board of each centre and de-identified data transmitted between centres. Individual patient consent was waived due to the retrospective nature of the study.

For each patient we collected the following data: age, gender, preoperative CT, FDG-PET and SRS findings, preoperative diagnosis, type of operation, pathological reports and follow-up. The decision to perform a lymph node sampling or a complete lymph node dissection was based on the surgeon’s usual practice. However, when lymph nodal sampling was performed a minimum of three stations were sampled based on the primary tumour location. Targeted biopsies based on the FDG_PET scan were not performed. The pathological results obtained from lymph node sampling and complete lymphadenectomy were combined for analysis based upon a preliminary analysis that showed the extent of lymph node dissection did not affect survival. Standardized definitions for each data point were decided a priori based on previous literature and distributed to each centre for use. Where possible, imaging studies were reviewed by each local investigator to confirm the findings in the radiologic reports.

The CT, FDG-PET and SRS results were evaluated for the presence of hilar and mediastinal lymph node metastasis. The presence of a lymph node metastasis was defined using the size criterion of 1 cm or greater in any axis on CT, whereas a positive node on FDG-PET was any node with an SUV > 2.0 and on SRS a positive scan was defined as the presence of uptake. No uptake or ‘negative’ noted in the report was defined as negative.

The imaging results were compared to the pathological findings in order to determine sensitivity, specificity, positive predictive value, negative predictive value and diagnostic accuracy of CT, FDG-PET and SRS in lymph node staging.

Sensitivity and specificity of FDG-PET and SRS were compared separately to those of CT in order to assess their utility in lymph node staging. Moreover, the test performances of CT, FDG-PET and SRS of patients with a TC were compared to those of patients with AC.

Continuous data were reported as median with interquartile range (IQR). Categorical and count data were presented as frequencies and percentages. Sensitivity and specificity between CT vs FDG-PET and CT vs SRS were compared using McNemar’s test if the groups were paired or by Chi-square test if the groups contained different patients. Overall survival was defined as the time interval in months from the date of surgery until the last follow-up or death and it was analysed by the Kaplan–Meier method. Disease-free interval was calculated in months from the date of surgery to the first recurrence of disease. A P-value <0.05 was considered significant. Statistical analyses were undertaken using SPSS 24.0 software (IBM Corp, Armonk, NY).

RESULTS

From 2000 to 2015, 380 patients underwent lung surgery for pulmonary carcinoid tumour. A total of 78 patients who had no lymph nodes sampled were excluded from the study leaving 302 patients for analysis. There were 106 (35%) men and the median age was 58 (IQR 47–68) years.

During the preoperative period, all patients underwent CT with a median period of 47 (IQR 27–72) days between CT and surgery. CT detected hilar lymph node metastasis in 31 cases and mediastinal lymph node metastasis in 36. FDG-PET was performed in 147 patients, with a median period of 40 (IQR 23–75) days before surgery. Positive hilar and mediastinal lymph node uptake was reported in 18 and 16 patients respectively. SRS was performed in 53 patients preoperatively with a median period of 27 (IQR 8–60) days between SRS and surgery. SRS identified positive hilar and mediastinal lymph node in 4 cases each.

At surgery, the primary tumour was surgically resected in all cases. There were 181 (60%) patients who underwent hilar-mediastinal lymph node sampling and 121 (40%) underwent complete lymph node dissection. Lymph node sampling and complete lymph node dissection were performed respectively in 60% and 40% of cases among patients who underwent CT, in 65% and 35% among those who underwent FDG-PET and in 72% and 28% among those who underwent SRS, without significant differences among the three groups (P = 0.26). The details about the surgical procedures are reported in Table 1.

Table 1:

Patients surgical procedure and pathological findings

Baseline characteristics
Sample size, n	302
Surgical procedure, n (%)
Lobectomy/sleeve lobectomy	211 (70)
Wedge resection/segmentectomy	44 (14)
Bilobectomy/sleeve bilobectomy	26 (9)
Pneumonectomy/sleeve pneumonectomy	13 (4)
Bronchial sleeve resection	8 (3)
Lymphadenectomy, n (%)
Sampling	181 (60)
Complete	121 (40)
Histology, n (%)
Typical carcinoid	244 (81)
Atypical carcinoid	58 (19)
Median tumour size, cm (IQR)	2.0 (1.4–3.2)
pStage, n (%)
I	218 (72)
II	50 (16)
III	32 (11)
IV	2 (1)

Pathologically, there were 244 (81%) cases of TC and 58 (19%) of AC. The median tumour size was 2.0 cm (IQR 1.4–3.2 cm). There were no hilar and mediastinal lymph node metastasis in 247 (82%) patients, 46 (15%) patients had ipsilateral hilar lymph node involvement (N1), 24 (8%) had mediastinal lymph node disease (N2) and 2 (1%) had contralateral hilar-mediastinal lymph node metastasis. The pathological findings data are reported in Table 1.

Sensitivity, specificity, positive predictive value, negative predictive value and accuracy of CT, FDG-PET and of SRS in detecting N1 and N2 disease are reported in Tables 2 and 3. No significant differences were observed when comparing sensitivity and specificity of CT to FDG-PET (P = 1.0 and P = 0.37 for N1 and N2 disease respectively) and to SRS (P = 0.47 and P = 0.35 for N1 and N2 disease respectively).

Table 2:

Comparison between imaging and pathological findings for N1 lymph node stations

	CT scana	FDG-PETb	SRSc
Imaging-pathology findings and test characteristics	(n = 291)	(n = 144)	(n = 49)
Imaging positive-pathology positive, n	15	9	2
Imaging positive-pathology negative, n	16	9	2
Imaging negative-pathology positive, n	30	15	7
Imaging negative-pathology negative, n	230	111	38
Sensitivity, % (95% CI)	33  (20–49)	38  (20–59)	22  (4–60)
Specificity, % (95% CI)	93  (89–96)	93  (86–96)	95  (82–99)
Positive predictive value, % (95% CI)	48  (31–67)	50  (27–73)	50  (9–91)
Negative predictive value, % (95% CI)	88  (84–92)	88  (81–93)	84  (70–93)
Accuracy, % (95% CI)	84  (80–88)	82  (71–93)	82  (71–93)

CI: confidence interval.

a

Data not available in 11 patients.

b

Data not available in 3 patients.

c

Data not available in 4 patients.

Table 3:

Comparison between imaging and pathological findings for N2 lymph node stations

	CT scana	FDG-PETb	SRSc
Imaging-pathology findings and test characteristics	(n = 291)	(n = 144)	(n = 49)
Imaging positive-pathology positive, n	10	9	3
Imaging positive-pathology negative, n	26	7	1
Imaging negative-pathology positive, n	12	6	6
Imaging negative-pathology negative, n	243	122	39
Sensitivity, % (95% CI)	46 (25–67)	60 (33–83)	33 (9–69)
Specificity, % (95% CI)	90 (86–93)	95 (89–98)	98 (85–100)
Positive predictive value, % (95% CI)	28 (15–45)	56 (31–79)	75 (22–99)
Negative predictive value, % (95% CI)	95 (92–97)	95 (90–98)	87 (73–94)
Accuracy, % (95% CI)	87 (83–91)	91 (86–96)	86 (76–96)

CI: confidence interval.

a

Data not available in 11 patients.

b

Data not available in 3 patients.

c

Data not available in 4 patients.

The same indices were calculated for TC and AC and reported in Tables 4 and 5. For both TC and AC, the sensitivity and specificity of CT in detecting N1 and N2 disease are similar to FDG-PET (TC: P = 1.0 and P = 0.30 for N1 and N2 disease respectively; AC: P = 1.0 and P = 1.0 for N1 and N2 disease respectively) and to SRS (TC: P = 0.54 and P = 0.39 for N1 and N2 disease respectively; AC: P = 0.85 and P = 0.85 for N1 and N2 disease respectively).

Table 4:

Comparison between imaging and pathological findings for N1 lymph node stations divided in typical and atypical carcinoid

	Typical carcinoid			Atypical carcinoid
	CT scana	FDG-PETb	SRSc	CT scanc	FDG-PET	SRSc
Imaging-pathology findings and test characteristics	(n = 235)	(n = 107)	(n = 35)	(n = 56)	(n = 37)	(n = 14)
Imaging positive-pathology positive, n	7	3	0	8	6	2
Imaging positive-pathology negative, n	12	6	1	4	3	1
Imaging negative-pathology positive, n	22	8	3	8	7	4
Imaging negative-pathology negative, n	194	90	31	36	21	7
Sensitivity, %  (95% CI)	24   (11–44)	27  (7–61)	–	50  (26–74)	46  (20–74)	33  (6–76)
Specificity, %  (95% CI)	94  (90–97)	94  (86–97)	97  (82–100)	90  (75–97)	88  (67–97)	88  (47–99)
Positive predictive value, %  (95% CI)	37  (17–61)	33  (9–69)	–	67  (35–89)	67  (31–91)	67  (13–98)
Negative predictive value, %  (95% CI)	90  (85–93)	92  (84–96)	91  (75–98)	82  (67–91)	75  (55–89)	64  (32–88)
Accuracy, %  (95% CI)	86  (82–90)	87  (81–93)	86  (75–98)	79  (68–90)	73  (59–87)	64  (42–89)

CI: confidence interval.

a

Data not available in 9 patients.

b

Data not available in 3 patients.

c

Data not available in 2 patients.

Table 5:

Comparison between imaging and pathological findings for N2 lymph node stations divided in typical and atypical carcinoid

	Typical carcinoid			Atypical carcinoid
	CT scana	FDG-PETb	SRSc	CT scanc	FDG-PET	SRSc
Imaging-pathology findings and test characteristics	(n = 235)	(n = 107)	(n = 35)	(n = 56)	(n = 37)	(n = 14)
Imaging positive-pathology positive, n	4	2	1	6	7	2
Imaging positive-pathology negative, n	23	5	0	3	2	1
Imaging negative-pathology positive, n	7	2	1	5	4	5
Imaging negative-pathology negative, n	201	98	33	42	24	6
Sensitivity, %  (95% CI)	36  (12–68)	50  (9–91)	50  (3–97)	55  (25–82)	64  (32–88)	29  (5–70)
Specificity, %  (95% CI)	90  (85–93)	95  (88–98)	100  (87–100)	93  (81–98)	92  (73–99)	86  (42–99)
Positive predictive value, %  (95% CI)	15  (5–35)	29  (5–70)	100  (5–100)	67  (31–91)	78  (40–96)	67  (13–98)
Negative predictive value, %  (95% CI)	97  (93–99)	98  (92–100)	97  (83–100)	89  (76–96)	86  (66–95)	55  (15–82)
Accuracy, %  (95% CI)	87  (83–91)	93  (88–98)	97  (91–100)	86  (77–95)	84  (72–96)	57  (31–83)

CI: confidence interval.

a

Data not available in 9 patients.

b

Data not available in 3 patients.

c

Data not available in 2 patients.

CT sensitivity and specificity in lymph node staging were similar between TC and AC (sensitivity: P = 0.08 and P = 0.39 for N1 and N2 disease respectively; specificity: P = 0.33 and P = 0.64 for N1 and N2 disease respectively). Conversely, CT negative predictive value is higher in detecting both N1 and N2 disease in TC than in AC (90% vs 82% for N1 disease; 97% vs 89% for N2 disease) but without a significant difference between the two histologies (P = 0.13 and P = 0.08 for N1 and N2 disease respectively).

Likewise, FDG-PET sensitivity and specificity in detecting N1 and N2 metastasis are similar between TC and AC (sensitivity: P = 0.60 and P = 0.91 for N1 and N2 disease respectively; specificity: P = 0.54 and P = 0.93 for N1 and N2 disease respectively). Nevertheless the FDG-PET negative predictive value in detecting both N1 and N2 disease is significantly higher in TC than in AC (N1: 92% vs 75%, P = 0.04; N2: 98% vs 86%, P = 0.03).

SRS specificity in lymph node staging is similar between TC and AC (P = 0.86 and P = 0.39 for N1 and N2 disease respectively) but negative predictive value is higher in TC than in AC, with a significant difference regarding N2 disease detection (N1: 92% vs 75%, P = 0.09; N2: 97% vs 55%, P = 0.002).

As for the presence of distant metastasis, 10/291 patients had extra-pulmonary findings suspicious for metastases on CT: 6 hepatic, 2 osseous, 1 pulmonary and 1 thyroid. These were all verified to be negative prior to pulmonary resection. Three of these patients however later developed metastatic disease within 13 months of surgery (IQR: 7–13). Comparatively, FDG-PET suggested the presence of suspicious bone metastasis in 3/144 patients (FDG-PET results are not available for 3 patients) and only in one case was it confirmed. SRS found suspicious metastatic lesions in 5/49 patients (SRS results are not available for 4 patients). The suspicious lesions involved liver in 3 cases, and thyroid in 1 (different from the thyroid identified at CT) and only this last one was confirmed to be a metastasis.

Follow-up was obtained for 295 patients. The median follow-up was 51 (IQR 18–97) months. The disease specific 5-year survival rate was 88.9% (95% CI 84.5–93.3%). During follow-up 36 (12%) patients recurred with a median disease free interval of 23 (IQR 10–50) months. The recurrence was local in 3 (1%) cases, regional (ipsilateral lung or ipsilateral hilar-mediastinal lymph nodes) in 12 (4%) and systemic in 21 (7%). In one case the tumour recurred systemically in bone and liver within three months of surgery. This patient had an AC clinically staged as IIIA with N2 disease. The preoperative CT identified a suspicious lesion in the liver that was negative at SRS and FDG-PET and so considered clinical M0 at that time.

Survival was also compared between patients who underwent CT preoperatively alone with patients that had not only CT but also the additional FDG-PET and/or SRS. No differences were identified in terms of disease specific survival (5-year survival rate: 94% vs 89% respectively; P = 0.07) and systemic recurrence rate (5.7% vs 6.2% respectively; P = 0.86) between the two groups.

DISCUSSION

In this study, the primary finding is that CT, FDG-PET and SRS have similar performance characterized by a low sensitivity but a high specificity, and a good negative predictive value in detecting hilar and mediastinal lymph node metastasis from pulmonary carcinoid. These results are similar to those reported in the literature [3–6] and call into question the need for additional imaging studies beyond a CT scan in the hilar and mediastinal staging of pulmonary carcinoids.

Although FDG-PET is commonly used for lung cancer staging, its value in the preoperative management of pulmonary carcinoid tumours remains controversial and in the present study was only performed in less than 50% of the cases. Recent studies by Pattenden et al. and Tatci et al. also reported a low sensitivity, a good specificity and a high negative predictive value of FDG-PET in the detection of nodal involvement suggesting that FDG-PET is not a reliable tool in the lymph node staging of lung carcinoids and that at a minimum nodal sampling is required [3, 4]. Conversely Gasparri et al. concluded in favour of the use of FDG-PET in the preoperative work-up of patients with suspected pulmonary carcinoid tumours [5]. However, it appears that these authors based their conclusions on the ability of FDG-PET to assess the primary tumour and not necessarily the nodal involvement. They did not specifically report the FDG-PET sensitivity, specificity and negative predictive value in detecting lymph node metastasis, however these indices can be calculated from their published data and were respectively 13%, 94% and 84%. These results are similar to those of our study, and underscore that FDG-PET may not provide additional value in terms of lymph node staging beyond a CT scan [5].

SRS is used much less commonly that CT or FDG-PET in evaluating pulmonary carcinoids but its performance is also similar to that of CT and FDG-PET. This suggests that SRS does not add further information about lymph node staging beyond the other two tests. Our conclusion differs from that of Kuyumcu et al. who stated that SRS was useful when added to FDG-PET because SRS confirmed the absence of nodal disease after a false positive FDG-PET scan. However, a counter argument to this conclusion is that a lymph node sampling will ultimately confirm or refute both the FDG-PET finding as well as the SRS finding [6]. Some clinicians point to the fact that SRS is considered a first line imaging technique in gastrointestinal neuroendocrine tumours providing whole-body information on local and distant disease and they consequently apply SRS to pulmonary carcinoids. However, unlike gastrointestinal carcinoid tumours, pulmonary carcinoid seems to have an arbitrary lack of somatostatin receptors leading to SRS false negative results [6–8]. This fact, combined with the lack of difference of SRS lymph node staging accuracy compared to CT or FDG-PET, suggests that SRS should not be utilized in the management of pulmonary carcinoids.

One rationale for using these additional imaging studies is that it would rule out nodal metastasis particularly when AC is encountered. However in our study and in Tatci et al., FDG-PET performed slightly better in excluding the presence of a lymph node involvement in TC than in AC (specificity and negative predictive value are respectively 79% and 91% for AC and 85% and 100% for TC) [4]. This suggests that even when a staging study suggests the absence of nodal metastases some form of nodal sampling should be undertaken especially when an AC is confirmed. Even in the presence of clinical N2 nodal metastases especially for TC, we favour surgical resection with lymphadenectomy since the 5-year survival after surgery in these patient’s is still an acceptable 70% or greater and induction therapy remains controversial and investigational. Moreover, we believe that for optimal staging a complete lymph node dissection should be performed for AC because of the higher false negative nodal imaging rate. As well, at least lymph node sampling is recommended for TC, whose false negative lymph node rate is lower, but not zero.

Given that FDG-PET and SRS do not provide additional information in terms of lymph nodal staging beyond that of CT, the question remains what to do when faced with a solitary pulmonary nodule suspected to be a carcinoid. Our approach is to confirm the histology first either by needle biopsy or bronchoscopic means and then proceed to resection with appropriate nodal sampling or dissection, without additional nuclear imaging/staging. We recognize however that often a knee-jerk reaction leads to order the FDG-PET scan earlier in the evaluation of such nodules on the presumption that it is a non-small-cell carcinoma. It may be difficult to stop this reflexive desire to obtain a FDG-PET scan. We however feel strongly in forgoing the SRS scan in pulmonary carcinoid staging once histological confirmation has been obtained.

Our study focused on the utility of FDG-PET and SRS in detecting nodal metastasis. We cannot comment about the role of these tests in the evaluation of distant metastasis since only patients who underwent surgical treatment for localized disease were included. However, we can speculate that CT staging is also adequate for assessing distant sites when we evaluate whether the studied patients developed distant metastasis on long term follow-up. In fact, there was no difference in disease specific survival between patients who underwent only preoperative CT and patients who had additional staging with FDG-PET and/or SRS. Accordingly, we speculate that the performance of FDG-PET and SRS tests in detecting distant metastasis is similar to that of CT.

This study has some limitations mostly based on its multicentre and retrospective nature. First, direct viewing of all imaging studies was not feasible leaving us to rely on reported findings. Second, FDG-PET, SRS and CT were performed in different institutions with different protocols, which may influence the results. Third, a centralized reviewed process for both pathology and radiology was not available. However, all results were reviewed in each centre in order to better define the preoperative lymph node staging for each imaging method, thus providing more granular data than an administrative data study. However, these three limitations can also be seen as strengths because it makes the study more generalizable since the performance characteristics were derived from multiple centres and physicians. Lastly, all three tests were not used on every single patient.

CONCLUSION

CT, FDG-PET and SRS all have similar performance in assessing pulmonary carcinoid lymph node staging. These findings suggest that additional nuclear imaging beyond CT is not required in the evaluation of otherwise potential candidates for resection. Despite a commendable negative predictive values of 88% and 95% for N1 and N2 involvement respectively, we however recommend that lymphadenectomy or nodal sampling be completed at resection for complete staging.

Acknowledgements

The authors would like to acknowledge the contribution of Ms Sandra Blitz for her statistical input and advice.

Conflict of interest: none declared.

Author notes

†Presented at the 24th European Conference on General Thoracic Surgery, Naples, Italy, 29 May-1 June 2016.