NEP-Score Thresholds Predict Survival of Patients With Bronchial Carcinoids

Survival prognostic markers are extremely needed to better define therapeutic strategies in patients with bronchial carcinoids (BC). We aim to verify the applicability of the NEP-Score in a homogeneous BC cohort and identify a derivative prognostic marker, the NEP-Score at diagnosis (NEP-D) that does not consider new metastases during follow-up. Sixty-four patients (38 females, and 26 males, mean age at diagnosis 58.9 ± 1.7 years) with BC were retrospectively evaluated. NEP-Score was calculated at the end of follow-up (NEP-T). A derivative score, the NEP-Score at diagnosis (NEP-D) that does not consider new metastases during follow-up, was then assessed. Patients were subdivided according to their living status at the end of follow-up. A NEP-Score threshold was investigated to predict survival. Mean NEP-T and mean NEP-D were significantly lower in live patients at end of follow-up. A NEP-T cut-off >138 significantly predicts survival. Atypical BC relapsed more frequently than Typical BC. Male gender and previous malignancy were negative prognostic factors for survival. We confirmed NEP-Score applicability in BC and NEP-D utility, being the latter a simple, quick, and cheap prognostic score that can help clinicians in decision making. The identified NEP-D threshold can predict NEN aggressiveness and may be used to define the best personalized therapeutic strategy. In this context, a validation study is needed.


INTRODUCTION
Bronchial carcinoids (BC) are uncommon bronchial/pulmonary tumors characterized by a wide spectrum of clinical behavior, representing 20%-30% of all neuroendocrine neoplasms (NEN), whose incidence rates range from 0.2 to 2/100.000 people/year, and most series suggest a higher incidence in women as compared to men and in Caucasian as compared to black patients (1)(2)(3)(4)(5)(6). In the United States Surveillance, Epidemiology, and End Results database, bronchial NEN annual incidence between 2000 and 2012 was 1.49 per 100,000 population (6). Several reports suggest that BC incidence is increasing over time (2,3,6); this may be at least partly related to the increased use of advanced medical imaging techniques that detect a higher number of asymptomatic tumors. Mean age at diagnosis for Typical BC (TBC) is 45 years, while in many series, patients with Atypical BC (ABC) are~10 years older (7,8), possibly influencing prognosis. BC management is mainly influenced by tumor differentiation and Ki-67 (9)(10)(11)(12)(13)(14), but markers of clinical outcomes that could predict patient survival are still lacking (15)(16)(17)(18)(19). Despite improvements in prognostic grading and staging systems, the challenge to predict BC patient's outcome is difficult. This is particularly true for patients with ABC, since they often have a worse prognosis with a greater tendency to metastasize and recur locally (20), with more frequent distant metastases (liver or bone) more frequent as compared to local recurrence (20). In addition, nodal metastases have an adverse influence on ABC prognosis, that is worse as compared to TBC. On the contrary, TBC usually have an excellent prognosis following surgical resection and the prognostic impact of nodal involvement remains controversial. Overall, BC may display a wide spectrum of clinical behavior, ranging from indolent to aggressive, with few validated prognostic factors that could help clinicians to predict survival. Several putative markers have been considered, including bloodbased biomarkers, such as Chromogranin A [CgA] (21), circulating tumor cells and microRNAs (22), as well as tissue markers (23), but none has been fully validated, so far. Pusceddu and co-authors (24) developed a classification prognostic score for overall survival (OS) in patients with well differentiated NEN, named NEuroendocrine Prognostic Score (NEP-Score], that we recently validated in an independent cohort of entero-pancreatic NEN (25). In the study by Pusceddu (24), NEP-Score turned out to be useful to rank patients according to their mortality risk and to predict OS. The aim of the present study is to verify NEP-Score applicability in a homogenous BC cohort and to identify a derivate marker capable to predict patients' prognosis by taking into account clinical and pathological characteristics at diagnosis. Furthermore, we considered the impact of other prognostic factors, such as gender, differentiation, and previous malignancy.

Study Design
We retrospectively evaluated the NEP-Score in a series of BC patients referring to our center from 1992 to 2018. According to Pusceddu S. et al. (24), only patients with a >24 months followup with TBC or ABC with a Ki-67 index 0%-2% and 3%-20%, respectively, were included, while patients with poorlydifferentiated neuroendocrine carcinomas were excluded.

Patients
We collected data on 64 patients (38 females, 26 males; mean age at diagnosis 58.9 ± 1.7 years) including seven ABC and 57 TBC followed up in our center from 1992 to 2018. Patients were evaluated for the following characteristics to calculate the NEP-Score at the end of follow-up (NEP-T) which lasted 98.2±11.3 months: age, site of primary tumor, primary tumor surgery, symptoms, Ki-67, timing of metastases, assigning the respective scores [see Table 1]. A modified NEP-Score at diagnosis (NEP-D), which does not take into account the appearance of new metastases during follow-up, was then calculated. Patients were subdivided according to their vital status (alive or not) at the end of follow-up (EOF). We also considered the difference between NEP-T score and NEP-D score, indicated as Delta NEP, in order to appropriately evaluate progression during follow-up. We considered as positive a Delta NEP>0, and as negative a Delta NEP=0. Patients characteristics are displayed in Table 2. This study is in accordance with the principles set out in the Declaration of Helsinki, has been specifically approved by the Local Ethics Committee (Comitato Etico Indipendente di Area Vasta Emilia Centro, CE-AVEC, at the Policlinico S. Orsola-Malpighi in Bologna) and authorized by the General Director of the Azienda Ospedaliero Universitaria in Ferrara (protocol number CE-AVEC 238/2020/Oss/AOUFe). Each patient has been informed of the purpose and nature of all procedures used.

Statistical Evaluation
Categorical data were summarized using frequencies and percentages. The chi-square test was performed to evaluate the presence of statistically significant differences among the evaluated groups in terms of NEP-Score. The paired Student t-test was employed to compare mean NEP-D and NEP-T scores among groups. A p-value <0.05 was considered significant. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated for each identified NEP-D and NEP-T threshold. Statistical analysis was performed using GraphPad Prism version 6.01 for Windows, GraphPad Software, San Diego, California USA, www.graphpad.com.

NEP-D Score Calculation
A modified NEP-Score was then calculated, considered as the NEP-Score at diagnosis (NEP-D), which does not take into account the appearance of new metastases during follow-up. Mean NEP-D was 108.5 ± 2.6. Patients were subdivided according to whether they were alive or not at EOF. We found that mean NEP-D in live patients (105.2 ± 3) was lower as compared to DOC (117 ± 5.7) and DNEN patients (121 ± 7.1), but these differences were not statistically significant. Therefore, NEP-D does not seem to be useful as a prognostic score in BC in our settings.

Gender
We also considered a possible difference between genders. In males, mean NEP-T of the 15 patients (mean age at EOF = 65.7 ± 4.4 years) who were alive at EOF was significantly lower as compared to that of the 11 patients (mean age at EOF = 75.2 ± 3.3 years) who were dead at EOF (117.7 ± 9.4; p<0.01) ( Figure  1A). This difference was not correlated to aging, since mean age at EOF in the two groups was not significantly different. In females, mean NEP-T of the 33 patients (mean age at EOF= 64.7 ± 2.2 years) who were alive at EOF was similar to that of the four patients (mean age at EOF = 71.4 ± 5.3) who were dead at EOF (110.6 ± 5.2 vs. 117 ± 17; p = not significant) ( Figure 1B).
Comparing males and females (dead and alive at EOF), we found that mean NEP-T was significantly higher in males as compared  . Therefore, NEP-T mirrors NEN-related prognosis and is not useful as a prognostic score in DOC patients in our settings.
Concerning NEP-D, as shown in Figure 1A, in male patients mean NEP-D of the 15 patients (mean age at diagnosis = 54.4 ± 4.1 years) who were alive at EOF was significantly lower as compared to that of the 11 patients (mean age at diagnosis = 67.9 ± 3.6 years) who were dead at EOF (105 ± 5.3 vs. 123 ± 5.2; p<0.05). This difference could be correlated to aging, since mean age at diagnosis in the two groups was significantly different (p<0.05). As shown in Figure 1B, in female patients mean NEP-D of the 33 patients (mean age at diagnosis = 57.3 ± 2.2 years) who were alive at EOF was similar to that of the four patients (mean age at diagnosis = 63.7 ± 2.4 years) who were dead at EOF (105.1 ± 3.8 vs. 107.5 ± 7.5; p= not significant). Comparing males and females, we found that mean NEP-D was similar in all groups, taking into account the influence of aging.
Females had a better survival as compared to males, when taking into account only DNEN patients (Figure 2A). In addition, mean NEP-D was similar in males (112.8 ± 4.1) and in females (105.6 ± 3.4), while mean NEP-T was significantly higher in males as compared to females (135.6 ± 6.9 vs. 111.3 ± 4.8; p<0.005). Interestingly, mean Delta NEP was significantly higher in males as compared to females (22.7 ± 5.4 vs. 5.8 ± 2.9; p<0.005), suggesting that disease progression/recurrence is more likely to occur in males as compared to females. Indeed, a higher number of male patients displayed a positive Delta NEP as compared to females (p<0.01). Indeed, male patients relapsed more frequently as compared to females (46.1% vs. 10.5%; p<0.01), with a consequently higher NEN-related mortality in males as compared to females (30.7% vs. none). As shown in Figure 2B, Delta NEP positivity was associated with a worse survival. These data suggest that male gender may represent a negative prognostic factor in BC, in keeping with previous reports (15)(16)(17).

NEP-T Threshold
A NEP-T threshold was investigated to assess the reliability of this score to detect disease status. We found that a NEP-T score cut-off ≥138 could correctly differentiate patients alive from those dead at EOF. Indeed, 8 out of 18 patients with NEP-T ≥ 138 and none of 46 patients with NEP-T <138 were DNEN at EOF (Figure 3). A NEP-T≥ 138 threshold represents a cut-off allowing the best compromise between sensitivity (100%) and specificity (83.7%), with a PPV = 50%, a NPV = 100%, and accuracy = 85.9% to predict survival (dead vs. alive at EOF specifically for NEN). In addition, we found a higher recurrence rate in patients with NEP-T≥138 as compared to patients with NEP-T<138 (83.3% vs. 2.2%; p<0.01). Furthermore, patients with ABC were significantly more represented among patients with NEP-T≥138 as compared to patients with NEP-T<138 (23.5% vs. 4.2%; p<0.05).

NEP-D Threshold
A NEP-D score threshold was investigated to predict survival at diagnosis. We found that a value of NEP-D≥ 130 represents a cut-off which allows for the best compromise between sensitivity (50%) and specificity (69.4%), with a PPV = 21%, a NPV = 89.5%, and accuracy = 66.

Disease Recurrence
Recurrence was significantly more frequent among males vs. females (46% vs. 10.5%; p<0.05) and in patients who were dead as compared to those who were alive at EOF (87.5% vs. 19.5%; p<0.005). During follow-up, patients with ABC had a greater recurrence rate as compared to TBC (57.1% vs. 21%; p<0.05).

Second Malignancy
Eight out of 64 patients (12.5%) developed a second malignancy during follow-up: 1 DOC and 1 DNEN. Patients with second malignancy had a significantly higher mean NEP-T as compared to patients without a second malignancy (147 ± 11.2 vs. 116.9 ± 4.4; p<0.05).

DISCUSSION
Our results show that the NEP-Score, corresponding to the NEP-T score in our study, is a valuable prognostic tool in BC management, since higher NEP-T scores associated with a worse survival. Therefore, we provide further validation of the NEP-score in a homogenous cohort of BC, after demonstrating its applicability in a homogeneous cohort of stage IV enteropancreatic NEN (25). The possibility to identify a score that could predict BC prognosis is of great interest for clinicians in order to improve patient management. In these settings, several therapeutic approaches, including surgery, medical therapy with somatostatin analogs (SSA) (26) or everolimus (27)(28)(29) and PRRT (30) are indeed available. Scant data are available concerning efficacy of loco-regional approaches (chemoembolization, radioembolization, radiofrequency ablation) in BC (31). Since randomized studies are scanty and the disease is relatively uncommon the level of evidence is limited as compared to more common cancers (32). Therefore, an  optimal management of the disease is difficult to establish. Thus, it would be very important to identify those patients that may benefit of an aggressive treatment from those who could be spared from unnecessary therapy. Consolidated prognostic factors are represented by morphology (i.e., differentiation) and proliferation rate in terms of number of mitosis and proliferation index. Incomplete resection is the only widely accepted feature with negative prognostic significance. Prognostic models have been previously investigated to predict patients outcomes and tailor treatment. Filosso et al. (16) identified as negative prognostic factors male gender, age, previous malignancy, ECOG performance status, peripheral tumor, TNM stage. The presence of metastatic disease demonstrated to be the strongest negative prognostic factor (16,17). In a large, single-center series of metastatic BC, age, bone metastases, liver metastases and Ki-67 index significantly correlated with prognosis (33). Few studies investigated prognostic indexes capable of estimating survival in bronchial NEN. Pusceddu et al. identified the NEP-Score, that was found to be useful to stratify survival probability both in very heterogeneous patients' groups (24) and in a homogeneous cohort of stage IV entero-pancreatic NEN (25). Our present study demonstrated the validity of the NEP-Score in an independent series of 64 BC patients (7 ABC and 57 TBC) with very long follow-up. We indeed found that the NEP-T score, that takes into account the appearance of metastases during follow-up, was significantly associated with survival. In our series we evaluated the timing of metachronous metastases, 62.5% of which appeared after 24 months and 43.7% after 72 months, in keeping with previous findings (33) and supporting the indication for a long follow-up. We identified a NEP-T score threshold that could be useful to estimate OS. However, the limitation of this approach is represented by the need of a long follow-up of at least 24 months to calculate NEP-T score. On the other hand, a score that could be calculated at diagnosis, such as the NEP-D, could be more useful. Indeed, we found that a specific NEP-D threshold is associated with a higher OS, despite its reduced performance in predicting patients' outcome in our settings. In particular, the identified NEP-D threshold was not able to predict 5 years survival and PFS, probably due to the small sample size. On the other hand, we found that NEP-D was significantly higher in male (but not in female) patients who were dead at EOF, suggesting that this score could be useful to predict prognosis in males. In addition, males showed higher NEP-T scores, more frequent relapses, a greater increase in Delta NEP and higher mortality for BC as compared to females, further supporting male gender as an independent negative prognostic factor, in keeping with previous reports (18,19). Delta NEP score may offer more indications, since it measures the development of metastases, with a different score depending on the timing of metastases appearance (before or after 24 months from diagnosis). The evidence that this score is significantly higher in patients dead as compared to those alive at EOF suggests that development of metastases, specially extrathoracic, profoundly influences survival in BC patients. Finally, we found that relapse rate was 2.5 fold higher in ABC as compared to TBC, confirming tumor differentiation as an important prognostic factor, in agreement with previous reports (18,19).
Our study has some limitations, mainly represented by the small sample size and the low number of patients with ABC. Moreover, the influence of medical treatment outcome and therapeutic advances developed during the last years (29,32,33) were not taken into account. However, this first study opens the way to a validation study of the promising NEP-D score on several and larger cohorts of patients. Indeed, validation of this score in blarger BC cohorts, in a multicenter setting, is necessary in order to provide a better estimate of the NEP-D threshold, possibly contributing to improve patient's management and quality of life.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Local Ethics Committee (Comitato Etico Indipendente di Area Vasta Emilia Centro, CE-AVEC, at the Policlinico S. Orsola-Malpighi in Bologna). Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.