Bronchopulmonary Carcinoids causing Cushing Syndrome: Results from a Multicentric Study Suggesting a More Aggressive Behavior

Abstract Objective Cushing syndrome (CS) caused by bronchopulmonary carcinoids (BCs) is a very rare entity. The aim of this study was to revisit the features of a multicenter clinical series to identify significant prognostic factors. Methods From January 2002 to December 2013, the clinical and pathological data of 23 patients (treated in five different institutions) were retrospectively reviewed. Survival analysis was performed to explore the relative weight of potential prognostic factors. Results Median age and male/female ratio were 48 years and 14/9, respectively. Most (> 80%) of the patients presented with CS-related symptoms at diagnosis. Tumor location was peripheral in 13 patients (57%) and central in 10 (43%). All patients but two (treated with chemotherapy) underwent surgical resection with curative intent. Definitive cyto/histology was indicative of typical carcinoid (TC) in 16 cases (70%) and atypical carcinoid (AC) in 7 cases (30%). A complete remission of CS was obtained in 16 cases (70%). Lymph nodal involvement was detected in 11 cases (48%), with N2 disease occurring in 7 (∼ 30% of all cases). Four patients (22%) experienced a relapse of the disease after radical surgery. Overall 5-year survival (long-term survival, LTS) was 60%, better in TCs when compared with AC (LTS: 66 v s. 48%, p = 0.28). Log-rank analysis identified ECOG performance status, cTNM and cN staging, pTNM and pN staging, persistence of CS and relapses (local p = 0.006; distant p = 0.001) as significant prognostic factors in this cohort of patients. Conclusion BCs causing CS are characterized by a high rate of lymph–nodal involvement, a suboptimal prognosis (5-year survival = 60%, 66% in TCs) and a remarkable risk of relapse even after radical resection. Advanced stage, lymph–nodal involvement and the persisting of the CS after treatment correlate with a poor prognosis.


Introduction
Cushing syndrome (CS) is characterized by the overproduction of cortisol and has significant morbidity/increased mortality, if untreated. CS has been traditionally divided into ACTH-dependent CS and ACTH-independent CS. The first form (comprising $ 80% of cases) is caused by a corticotropic pituitary adenoma and rarely by ectopic ACTH production. 1 Critical illness is related to glucocorticoid excess and can occur in all cases of CS. 2 As widely reported, 1,2 ectopic ACTH secretion has been described in a wide array of neoplasms, the most prevalent being the bronchopulmonary carcinoid (BC, 1-5% associated with ectopic ACTH secretion, representing 1-10% of CSs 3-7 ). First described in 1928, 8 less than 100 cases of BCs causing CS have been reported in the English literature up to 2004 5 with only few series that describe more than five patients. 4,9,10 A meta-analysis performed in early 2000s was focused on 19 cases only with BC-related CS. 11 The clinical decision-making process in these challenging cases is based, so far, on the evidences stemming from series of case reports referring to single cases or few small retrospective series. 9 Moreover, the preoperative diagnostic examination and clinical investigation for ectopic ACTH sources may be often very challenging; thus, the final diagnosis of BCs associated with ACTH secretion may be significantly delayed from the clinical onset.
In addition, because of the scarcity of data, the biological behavior and the prognosis remain unknown and debated. Indeed, some authors 3,9,10 have suggested that ACTH secreting BCs represent a substantially more aggressive subtype of BCs with a high rate of mediastinal lymph node involvement at the moment of diagnosis and a suboptimal prognosis when compared with the other forms of BC. Therefore, with the aim of better defining the clinical presentation, histopathological characteristics, treatment modalities, and long-term survival results in patients with BCs causing CS, we collected and studied consistent and homogeneous large cohorts of patients. The outcome of this analysis is reported herein.

Materials and Methods
In the period between January 2002 and December 2013, a total of 23 patients were treated for BCs causing CS in the five Italian centers involved in this analysis and these formed the basis for this retrospective analysis. Before undertaking our data analysis, we obtained the Promoting Centre IRB approval (Protocol Number: 2014/6230). Data related to gender, age, signs and symptoms, laboratory tests, tumor location and stage, surgeons' notes, pathological features, postoperative therapy, recurrence patterns, and long-term follow-up (FU) were systematically reviewed and recorded (see ►Table 1).

Preoperative Evaluation
In all the patients, preoperative evaluation included history; physical examination; routine blood tests; standard chest Xray; electrocardiogram; spirometry; arterial gas analyses; pituitary computed tomography (CT)/magnetic resonance imaging (MRI); and bronchoscopy (lesions proxy to major airways).
In detail, the determination of ACTH serum values, cortisol serum values (at 8:00 AM), and urinary cortisol values were performed to confirm the clinical or subclinical suspicious of Cushing syndrome (CS) and the results are reported in ►Table 1.

Surgery
Surgery (lateral thoracotomy in all cases) was substantially homogeneous along with criteria to determine resection extension: (1) sublobar resection was indicated only in patients with poor pulmonary function; (2) lobectomy/pneumonectomy was indicated in the remaining cases; (3) systematic lymph-nodal dissection was performed in all the cases as described by Lardinois et al. 12 Pathology World Health Organization (WHO) 2004 lung cancer classification criteria 13 were used and TNM was assessed according to IASLC 2008. 14 As reported in ►Table 1, pNETs were classified as typical carcinoids (TCs ¼ 16 patients, 70%) or atypical carcinoids (ACs ¼ 7 patients, 30%). Flow-cytometric analysis was performed on fresh material. Immunohistochemistry was performed on formalin-fixed, paraffin-embedded tissue, using the streptavidin-biotin immunoperoxidase method, in addition to standard Hematoxylin and Eosin staining (three sections for each case). A panel of cell markers including chromogranin (always positive staining) was routinely performed in all centers; Synaptophysin was tested in 22 cases (all positive but one); thyroid transcription factor-1 in 18 cases (15 positive). Moreover, immunohistochemistry for the anti-ACTH antibody was performed in selected cases (unnecessary when post-op drop in serum ACTH occurs).

Postoperative Treatment and Follow-Up
Adjuvant chemotherapy and/or radiotherapy were performed under the care of referring oncologists; while a certain degree of homogeneity was present within the five centers, this is lost when the centers are compared among themselves. The clinical records of outpatient clinics and correspondence obtained information about the health status Conclusion BCs causing CS are characterized by a high rate of lymph-nodal involvement, a suboptimal prognosis (5-year survival ¼ 60%, 66% in TCs) and a remarkable risk of relapse even after radical resection. Advanced stage, lymph-nodal involvement and the persisting of the CS after treatment correlate with a poor prognosis. Thoracic

Results
Demographical, clinical, and pathological features are summarized in ►Table 1. Overall, 9 women and 14 men (median age ¼ 48 years) were included in the study. Tumor was found peripheral in 57% and central in 43% of subjects. Average size was 2.1 AE 1.2 cm. At diagnosis, 19 patients (83%) presented with symptoms and, in 13 cases (57%) such symptoms were related with CS. Laboratory test results are reported in details in ►Table 1. CT scan was performed in all cases and failed to visualize the neoplastic lesion in four cases (17.4%). 18 F-FDG PET/CT scan evaluation showed an increased uptake of the tracer in four of six cases (66.7%). The two false-negative cases were TCs. On the contrary, 68 Ga (68-Ga DOTA-peptide) revealed an uptake in 9 of 10 suspected lesions (90%) with only one false-negative case (AC). At clinical staging, most patients presented with stage I disease (65%) although about one-third (30%) had a (locally) advanced disease. Surgery (see ►Table 1) was indicated in 21 of 23 cases (91%) of which four sleeve lobectomy, two sublobar resections (both performed in patients with poor pulmonary function where the diagnosis of BC was already achieved during the preoperative setting), and one left pneumonectomy for an endobronchial mass of the left main bronchus (sleeve lobectomy technically unfeasible). Residual macroscopic neoplastic tissue (R2) at the level of mediastinum was left after surgical resection in a young patient (39 years) with a TC invading the supra-aortic vessels. In two other cases, the bronchial stump was found to be microscopically (R1) involved (frozen section not routinely performed). Postoperative complications occurred in five patients (23.8%). Definitive pathology (details in ►Table 1 and ►Fig. 1): TC in Overall, a CS's remission was experienced by 16 patients (70%) with a rapid decrease of the laboratory parameters, whereas in 7 patients (two of them underwent nonsurgical therapy) we observed the persistence of clinical symptoms related with CS. Finally, after surgery, a total of six patients underwent adjuvant therapy, whereas the majority of them (17 patients, 74%) did not undergo any post-op treatment.

Pattern of Recurrence and Long-Term Survival
During the FU, four patients who had undergone R0 resection experienced a recurrence of disease (22.2% of all R0 patients). Among these, in one case, the relapse was intrathoracic only, in two cases, it was at the same time locally and distantly located and, finally, in one patient, widespread metastases occurred-22 months-after surgery. CS relapsed synchronously with the radiological detection of relapse in two of them, whereas, unfortunately, the clinical information regarding other cases were not available (or not clearly interpretable) at the moment of preparing the present report.
Overall survival is detailed in ►Fig. 2. Considering the entire cohort of patients, the 5-year long-term survivals (Kaplan-Meier) was 60% (95% CI, 30-80), respectively. From the log-rank test (►Tables 2 and 3), it emerged that: ECOG PS, clinical, and pathological TNM staging and NÀstatus, persistence of CS and local and/or distant relapses were factors significantly associated to survival. Borderline significance was achieved with regards to tumor necrosis (p ¼ 0.06), vascular embolism (p ¼ 0.07), and resection status (p ¼ 0.07). ►Table 2 shows all prognostic factors examined. The Cox regression analysis (see ►Table 3) showed that the number of mitoses and the Proliferation Index ki-67% significantly increase the hazards of mortality by a factor of 1.36 (95% CI, 1.01-1.84; p ¼ 0.045) and by a factor of 1.42 (95% CI, 1.07-1.89; p ¼ 0.015), respectively.

Discussion
The differential diagnosis of CS and in particular the differentiation of pituitary Cushing disease from ectopic ACTH secreting neoplasms can be difficult. 15 Although BCs are the most frequent cause of CS resulting from ectopic ACTH secretion, 16 this clinical condition is only rarely reported and several critical points remain unresolved. Indeed, excluding the case reports, only very few studies on small series that describe more than five patients (summarized in ►Table 4) have been so far reported. 3,9,10,17 In this context, we have planned a multicentric retrospective study collecting a relatively large clinical series of patients affected by ACTHsecreting BCs aiming to better define the clinical presentation and to identify the main prognostic factors in such a rare entity. 18

Clinical Presentation
CS-associated BCs occur in relatively young people (mean age % 50 years). In most cases, patients with BCs do not have respiratory symptoms or signs, thus, making clinical diagnosis difficult 15 ; on the contrary, they often present at diagnosis with symptoms related with the ectopic increase of ACTH production and often occurring several months before diagnosis. In our series, approximately 83% (number 19 patients) presented with one or more symptoms (mostly related with CS disease).

Diagnosis
The localization of the source of ectopic ACTH production is challenging. In the study in by Ejaz et al, 19 the time interval from CS diagnosis to the detection of the source of ACTH secretion was variable (range, 0-118 months) and ranges from 17 to 54 months in other series (►Table 4).
As reported in a recent review by Kenchaiah and Hyer, 20 serial CT and MRI scans often fail to localize 33 to 44% of ectopic corticotropin-producing tumors, 8,10,11 this mostly explaining the delayed time from clinical CS presentation to final diagnosis.
In fact, although the lung represents the most likely organ to harbor an ectopic source of ACTH, small peripheral BCs may be "easily" missed on chest CT scan evaluations. Thus, a careful analysis of pulmonary parenchyma during CT and the adoption of modern multidetector high-resolution CT (2.5 mm slices) are essential and, thus, strongly recommended.
In our study, we have observed four cases (17.4%) of BCs causing CS where the standard radiological assessment failed to identify the neoplastic lesion (or the pulmonary lesion was only suspected to be malignant). In these cases, the radiometabolic evaluation (18-F FDG and 68-Ga DOTA-peptides PET/CT scans) was pivotal to correctly identify such tumors.
In fact, in line with some authors, 21 but also in disagreement with others, 9 we believe that somatostatin receptor imaging may have a complementary role with standard radiological imaging in localizing ectopic ACTH secretion sites, and more generally to identify BCs. 22,23 Nevertheless, despite PET-CT imaging with Ga-68 DOTA-peptides represents a promising diagnostic tool in this clinical scenario, the results are still preliminary and controversial. 24 However, even with modern techniques, a source of ectopic ACTH may stay undetected in approximately 9 to 10% of patients. 19 Such cases with "occult" ACTH secreting tumors remain challenging and may require repeated investigations over many years.

Oncological Considerations and Prognostic Determinants
One of the most debated issue concerns the biological behavior of BCs causing CS. Some authors 9,10 have reported en-hanced aggressiveness of these neoplasms when compared with that of nonsecreting BCs. Boddaert et al 9 have recently reported the long-term results of a cohort of 14 patients affected by BCs causing CS. They observed a high rate of lymph nodal involvement at diagnosis (significantly higher when compared with lymph nodal involvement reported in the literature for nonsecreting BCs), suggesting a more aggressive behavior of this entity when compared with hormonally quiescent BCs. Our data (47% of pNþ disease with 30% of pN2Àdisease) are substantially in line with these findings. When considering TCs only (16 patients), we have found a pNþ disease in six of them (37%) with four patients having a N2 disease (25%) and two patients N1 disease (12%). These results substantially differ from those of nonsecreting TCs reported in our previous study (observed a 11.6% of Nþ disease (N1Àstatus ¼ 11.6% and N2Àstatus ¼ 0%) in 163 surgically treated TCs 24 ). Moreover, what it is interesting to note is that the preoperative NÀstatus revealed only six patients (26%) with N1-2 involvement, which theoretically implies a certain degree of inaccuracy of pre-op staging assessment. Actually, reviewing those cases where lymph nodal involvement was "missed" at pre-op setting, the pathological involvement was labeled as "microscopic" disease (lymph nodes < 1 cm in all cases), this probably explaining the detection failure of the diagnostic methods. Concerning the prognosis, as suggested by Shrager et al, 10 these neoplastic entities seem to represent a biologically more aggressive subtype of the usual nonsecreting BC tumors. Taking into account the overall 5 years survival in TCs (66% in this study), this prognosis may be considered "suboptimal," if we compared this with LTS for nonsecreting TCs in our previously experience (LTS ¼ 98.6% according to Cardillo et al 25 ) and similarly (91%) according to Filosso et al 26 ). The recurrence rate (ranging from 7-43%, see ►Table 4) may represent (even more than 5 years' survival) a direct index of the biological aggressiveness of such entities and implies the need of a careful and long-term clinical and radiological FU. Therefore, while nonsecreting TCs could be considered as an indolent disease and surgical treatment often less aggressive (lungsparing surgery), in TCs (and more generally in all BCs) causing CS, the surgical treatment should be the same as for high-grade malignant tumors and should include anatomic resection and radical lymphadenectomy. When investigating the long-term survival with the aim of identifying clinical and pathological prognostic factors, we have found that a poor ECOG score, the advanced clinical and pathological TNM stage, the lymph nodal involvement, the Rþ status, the presence of vascular embolism (observed in 26% of our cases),  (7) 48 and other pathological features indicative of biological aggressiveness (number of mitoses, K i index, and tumor necrosis) were all factors associated with a poorer long-term survival (see ►Tables 2 and 3). Probably, because of the small sample size, we did not observe any significant difference according with histotype.
In addition, and more interestingly, we have noted that the persisting of the CS after treatment is strongly associated with a poor prognosis in this cohort of patients (5 years' survival: 0 vs. 81%, respectively; p ¼ 0.004). This issue, never investigated in other series, deserves few considerations. First, we may hypothesize that the postoperative clinical (or subclinical) persistence of CS could represent a "warning alarm" for the persistence of neoplastic tissue (even microscopically). In this setting, there are some reports 10,27 describing the persistence of CS after the first (intended radical) surgical resection of the ACTH-secreting BC where a redo surgical approach was needed to remove the remaining neoplastic tissue.
The second consideration (and also the logical consequence of the first) is that surgery in these selected cases should be as radical as possible. In this context, we have also adopted (and still recommend) an innovative technique of intraoperative radio-guided surgery 28 with the aim of improving the completeness of resection in such challenging cases. Thus, pulmonary anatomic resection and a systematic mediastinal lymph-nodal dissection 12 are mandatory in any ACTH-secreting BC. Finally, in the case of persisting CS, a diagnostic work-up examination should be directed to identify persisting ACTH secretion sources both in the chest and/ or in other sites, even adopting advanced radiometabolic techniques (i.e., the 68-Ga DOTA-peptide PET/CT scan) if not used at baseline evaluation.

Limitations and Strengths
This analysis has significant limitations in solving the uncertainties around the ACTH-secreting BCs. First, it suffers from the common bias of retrospective investigations because of the fact that the data were collected from different centers. This fact, that could slightly influence the availability (i.e., no complete data regarding the delayed time from clinical CS onset and pathological diagnosis) and validity of  the body of data itself, should be kept firmly in mind by the readers when considering the clinical implications deriving from the present analysis. Finally, we are conscious of the fact that collecting a multi-institutional data in a prospective fashion could ensure better homogeneity of clinical data, but considering the rarity of such neoplasm we have excluded this kind of approach when planning the study design. Therefore, despite the aforementioned weak points, this study has the remarkable merit to collect a relative large clinical series (the largest one reported till now in literature, to the best of our acknowledge) supporting the discussion on this unusual and controversial issue (not achievable when planning a monocentric retrospective study). Moreover, the homogeneity of the cohort and the accurate review of the pertinent literature are additional clear points of strength of this study.

Conclusion
Bronchopulmonary carcinoids (BCs) causing CS are characterized by a high rate of lymph-nodal involvement at diagnosis (N2 disease in 30% of patients), a suboptimal prognosis (5-year survival ¼ 60%, 66% in TCs) and a remarkable risk of relapse ($ 22%) even after radical resection. Despite these results need to be confirmed on larger clinical series of patients, we may assume that these entities could be considered as a biologically more aggressive variant when compared with the other nonsecreting BCs, characterized by a less indolent behavior. Therefore, surgery (when indicated and clinically feasible) should be as radical as possible (pulmonary anatomic resections and systematic mediastinal lymph-nodal dissection). Advanced TNM staging and NÀstatus, high number of mitosis, and K i -67% index, and the persisting of the CS after surgery correlate with a poor prognosis.