Prognostic value of inflammation-based markers in advanced or metastatic neuroendocrine tumours

Background The role of systemic inflammation–based markers remains uncertain in advanced or metastatic neuroendocrine tumours (nets). Methods Systemic inflammatory factors, such as levels of circulating white blood cells and other blood components, were combined to yield inflammation-based prognostic scores [high-sensitivity inflammation-based Glasgow prognostic score (hsgps), neutrophil:lymphocyte ratio (nlr), platelet:lymphocyte ratio (plr), high-sensitivity inflammation-based prognostic index (hspi), and prognostic nutritional index (pni)], whose individual values as prognostic markers were retrospectively determined. Univariate and multivariate analyses were used to examine the association of inflammatory markers with overall survival (os). Results The study included 135 patients. Univariate analysis revealed that elevated white blood cell count, elevated neutrophil count, low serum albumin, elevated high-sensitivity C-reactive protein, and elevated hspi, hsgps, and nlr scores were significantly associated with worse os. Multivariate analyses demonstrated that, apart from pathology grade and original site of the tumour, elevated hspi (p = 0.004) was an independent prognostic factor for worse os. Conclusions In the present study, elevated pretreatment hspi was observed to be an independent predictor of shorter os in patients with inoperable advanced or metastatic net. The hspi might thus provide additional guidance for therapeutic decision-making in such patients.


INTRODUCTION
Neuroendocrine tumours (nets) constitute a heterogeneous group of malignancies that originate from cells of the endocrine system, most commonly the gastrointestinal tract 1 .Because these tumours have been regarded as relatively rare, their biology and molecular characteristics, and the optimal treatment strategy for affected patients, are far from clear [2][3][4] .In patients with net, the clinical course varies from highly aggressive disease, with affected patients living only approximately 6 months and developing high-grade metastatic tumours, to more indolent processes, with affected patients having a median survival of approximately 20 years 5 .It is therefore becoming more important to identify effective prognostic factors to guide clinical treatment.
Tumour stage, tumour grade, and site of tumour origin are well established prognostic factors for patients with nets 6 .However, even within the same classification of those factors, response to treatment and survival vary from patient to patient 7 .Additionally, a growing body of evidence demonstrates that patient outcomes are also determined by the tumour microenvironment-the systemic inflammatory response in particular 8 .
Measurement of the systemic inflammatory response has been refined through the use of various indicators, including plasma C-reactive protein (crp) and serum albumin, and a combination of crp and albumin that is termed the Glasgow prognostic score (gps).Those indicators have been shown to be independent prognostic factors in colorectal, gastric, and renal cancer [9][10][11] .Apart from those markers, many studies have demonstrated that other hematologic components of the systemic inflammatory response and certain specific combinations-for example, neutrophil:lymphocyte ratio (nlr), platelet:lymphocyte ratio (plr), prognostic index (pi), and prognostic nutritional index (pni)-can also serve as prognostic factors and are associated with survival in cancer patients [10][11][12][13][14][15][16][17] .
However, the role of those inflammation-based markers remains uncertain in advanced or metastatic net.We therefore examined the value of those markers as prognostic factors and the extent to which they improve the prognostic classification of such patients.

Patients
Our study was approved by the Medical Ethics Committee of Peking University Cancer Hospital, Beijing, P.R.C., and was performed according to principles of the Declaration of Helsinki.All study participants gave written informed consent for the storage of their information in the hospital database and its use for future research at the time of follow-up ascertainment.
Detailed clinical data for patients treated at the Gastrointestinal Oncology Department of Peking University Cancer Hospital were recorded in a regularly updated electronic database.Eligibility criteria included patients with n pathology-confirmed inoperable locally advanced or metastatic net.n a history of systemic chemotherapy, somatostatin analog, or targeted therapy use.n a life expectancy of 3 months or more.
Patients who received adjuvant chemotherapy within 6 months of recurrence and those showing clinical evidence of infection or other inflammatory conditions (for example, connective tissue disorders, rheumatologic diseases, and vasculitis) were excluded from the study.

Prognostic Index
Data were gathered from medical records, including general patient demographics [sex, age, and Karnofsky performance status (kps)]; tumour characteristics (location, number of metastases, degree of differentiation, and expression of somatostatin receptor); and pretreatment laboratory tests relating to the potential prognostic factors [white blood cell (wbc), neutrophil, lymphocyte, and platelet counts; serum neuron-specific enolase (nse), lactate dehydrogenase (ldh), and albumin concentrations).In addition, information about treatment methods was also extracted.
High-sensitivity crp (hs-crp) in the plasma sample collected at the patient's first visit was tested by human antienzyme-linked immunosorbent assay kit (Beckman Coulter, Brea, CA, U.S.A.).The laboratory variables were analyzed as categorical variables based on standard thresholds.Dichotomization of the variables was based on the upper (wbc count, neutrophil count, ldh, nse, hs-crp) and lower (albumin concentration, lymphocyte count) range of normal measurements for the markers.The cut-off points for dichotomization of the nlr and plr were their median distributions, which accorded with the cut-off points used for advanced malignancies in published studies (nlr ranging from 2.5 to 5.0 [18][19][20][21] ; plr ranging from 150 to 300 14,19,20,22 ).The gps, pi, and pni were also determined using the combination of systematic inflammatory markers specified in published reports 9,[14][15][16]23 . Thesystemic inflammationbased prognostic scores of gps, nlr, plr, hs-pi, and pni were determined as described in detail in Table i.

Statistical Analysis
All analyses were performed using the IBM SPSS Statistics software application (version 21 for Windows: IBM, Armonk, NY, U.S.A.).Differences between the study groups were determined using two-sided t-tests for continuous variables and Pearson chi-square tests for categorical variables [age, sex, type of tumour, primary tumour site, kps, laboratory tests (including wbc, neutrophil, and lymphocyte counts; and nse, ldh, albumin, and hs-crp concentrations)].Survival duration was calculated from the date of first visit to the date of death.Survival probabilities were compared for various categories of interest using the Kaplan-Meier method with log-rank test.A multivariate analysis of survival used the Cox proportional hazards model, adjusted for factors that were identified as significant (p < 0.05) on univariate analysis.All p values were two-sided, and p < 0.05 was considered significant.

Univariate Survival Analysis
The univariate analyses demonstrated that sex, tumour grade, original tumour site, presence of metastasis, number of metastases, body mass index, serum albumin, ldh, nse, hs-crp, wbc count, neutrophil count, lymphocyte count, hs-pi, hs-gps, and nlr were significantly associated with os (all p < 0.05).No significant correlation of os with age, kps, carcinoid syndrome, metastasis or not, somatostatin receptor scintigraphy status, lymphocyte count, pni, or plr was observed (all p > 0.05, Table ii).

Multivariate Survival Analysis
The variables sex, tumour grade, original tumour site, kps, presence of metastasis, number of metastases, body mass index, serum albumin, ldh, nse, hs-crp, wbc count, neutrophil count, lymphocyte count, hs-pi, gps, and nlr   were included in the multivariate analyses.The results demonstrated that pathology grade (p < 0.001), original tumour site (p = 0.01), and hs-pi (p = 0.004) were independent prognostic factors for survival (Table iii).Figure 1 shows the survival curves for patients by original tumour site, pathology grade, and hs-pi.
Patients with a higher hs-pi score tended to have more severe disease and worse overall condition, which could be associated with worse outcomes (Table iv).

DISCUSSION
Study findings suggest that systemic inflammation-based scores-and, in particular, an elevated pretreatment hs-pi score-are independent predictors of shorter os in patients with inoperable advanced or metastatic net.The hazard ratio for death was elevated by a factor of 1.5 for the hs-pi 1 group and by a factor of 4.75 for the hs-pi 2 group compared with the hs-pi 0 group.Despite advances in the accuracy of clinical staging, established prognostic factors are of limited prognostic value in this disease because of the rarity of this tumour type and the tremendous patient heterogeneity 24 .The 2010 World Health Organization classification of tumours of the gastrointestinal tract, liver, and pancreas, a grading scheme for nets of the digestive tract that is based on expert consensus and endorsed by the European Neuroendocrine Tumor Society 25 , has resulted in improvements in tumour prognostication, treatment planning, and comparison of data from different institutions 26 .However, many studies support the concept that the current World Health Organization G3 category is heterogeneous 27,28 .Tumours at the lower end of the G3 range are, in fact, well-differentiated nets with an elevated proliferative rate (that is, high-grade, well-differentiated nets), and prognosis appears to be   significantly better for those patients than for patients with poorly differentiated high-grade tumours 29 .The current World Health Organization high-grade neuroendocrine carcinoma category might therefore have to be refined and a new prognostic index developed 4 .Additionally, chromogranin A is the most frequently used noninvasive serum marker, especially in the management of patients with well-differentiated nets.However, it has some limitations, given that various assays are available, and international standardization is lacking.In addition, elevated chromogranin A can be caused by renal or liver failure and the use of proton pump inhibitors 30,31 .In patients with net, nse is elevated and correlates with tumour size 32,33 .It has been considered to be a useful marker for follow-up in several studies of net [34][35][36] and a generic marker for both neurons and neuroendocrine carcinoma.Although it has high sensitivity, its specificity is low.Similarly, elevated ldh is considered to be an adverse prognostic factor in various solid tumours, including nets [37][38][39] .In our analysis, abnormally increased levels of nse and ldh also predicted worse outcome in univariate analyses (p < 0.001), but lost statistical significance in Cox proportional hazards regression modelling, appearing to be less robust as prognostic markers in advanced or metastatic nets.More efficient factors therefore have to be uncovered.It is now becoming clear that the tumour microenvironment, which is largely orchestrated by inflammatory cells, is an indispensable participant in the neoplastic process.Many blood components, including acute-phase crp [40][41][42][43] , lymphocytes [44][45][46] , wbcs 47,48 , and neutrophils [49][50][51] , have been identified as markers that reflect the systemic inflammatory response.Moreover, to further refine prognostic accuracy, a variety of indices based on a combination of various inflammatory markers or a combination of inflammatory factors and albumin or platelet count have been proposed.Evidence has shown that systematic inflammatory factors such as gps, pi, pni, nlr, and plr provide superior prognostic value for cancers of the lung, breast, colorectum, and stomach 9,13,15,19 .In the report by Salman et al. 20 , it was also demonstrated that nlr and plr can serve as factors to reliably predict survival in gastroenteropancreatic nets.However, that study focused mainly on the roles of nlr and plr in gastroenteropancreatic nets.Because the relationship between crp-based systemic inflammation-related prognostic scores and advanced or metastatic nets has not been examined, we further considered the individual markers of crp, lymphocyte count, wbc count, neutrophil count, and the combined markers of gps, hs-pi, pni, nlr, and plr, analyzing their correlation with os and clinicopathologic parameters.To our knowledge, the present study is the first to investigate the roles of hs-pi, pni, and gps in nets.In addition to the traditional clinicopathologic parameters such as grade and tumour location, hs-pi was also found to be an independent predictor of reduced survival in patients with advanced or metastatic net, and it was a predictor superior to gps, nlr, plr, and pni.
Being a combination of plasma crp and wbc count, a hs-pi score that is elevated at the time of net diagnosis is independently associated with shorter os in advanced or metastatic net.A high hs-pi score indicates elevated crp and an increased number of wbcs.Those increases reflect imbalance in the immune response, which impairs normal antitumour functions.White blood cells are described as main sources of the inflammatory response and important participants in the production of circulating angiogenic growth factors that promote tumour progression.The acute-phase protein crp is produced by hepatocytes, predominantly under the control of circulating interleukin 6.Some cancer cell lines produce interleukin 6, which is strongly associated with serum crp.Increased serum crp has been recognized to be a prognostic factor for poor outcome in several malignant tumours, including nets.In a prospective cohort study, Siemes et al. 52 hypothesized that increased serum crp and crp gene variations are associated with an altered risk of colorectal, lung, breast, and prostate cancers.In addition, crp modulates both innate and adaptive immunity.Because the peripheral wbc count and hs-crp test are convenient, hs-pi can be considered to be a useful marker for predicting immune status in patients with advanced or metastatic net.
As part of our study, we considered the relationships of hs-pi with other clinicopathologic parameters, finding that hs-pi is not only associated with hs-crp, wbc count, and other crp-based inflammation-related scores (such as gps), but also with other systemic factors-for example, neutrophil count, pni, and nlr.Furthermore, a higher hs-pi score was relevant to the biologic characteristics of the tumour, such as poorer differentiation, and was found more often in male patients than in female patients.Moreover, an elevated hs-pi was associated with higher levels of nse and ldh, both of which are associated with worse outcomes.The statistically significant correlation between hs-pi and those related factors further suggests that hs-pi could be considered to be a reliable prognostic marker in advanced or metastatic net.

CONCLUSIONS
In summary, based on widely available, cost-effective, and easy-to-perform measurements, we developed a novel hs-pi.Using serum crp and the wbc count, the hs-pi identifies 3 subgroups of patients with advanced or metastatic net who have distinct prognoses.This newly proposed hs-pi might improve the accuracy of survival predictions in patients with advanced or metastatic net.It might also provide information that can complement other prognostic models, such as those based on gene profiling, potentially aiding in treatment decision-making and influencing a revised staging system.
Table ii details the patient characteristics.

TABLE III
Multivariate analyses of overall survival in 96 patients

TABLE IV
Correlation between the high-sensitivity inflammationbased prognostic index (hs-PI) and clinicopathologic parameters in 96 patients