Plasma Chromogranin A as Marker for Survival in Patients With Metastatic Endocrine Gastroenteropancreatic Tumors

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6: Plasma Chromogranin A as Marker for Survival in Patients With Metastatic Endocrine Gastroenteropancreatic Tumors RUDOLF ARNOLD,* ALEXANDRA WILKE,* ANJA RINKE,* CHRISTINA MAYER,* PETER HERBERT KANN,* KLAUS JOCHEN KLOSE, ANDRÉ SCHERAG,, MAIK HAHMANN, HANS HELGE MÜLLER, and PETER BARTH** *Department of Internal Medicine, Division of Gastroenterology and Endocrinology, Philipps University, Marburg; Department of Radiology, Philipps University, Marburg; Institute of Medical Biometry and Epidemiology, Philipps University, Marburg; Institute of Medical Informatics, Biometry and Epidemiology, University of Duisburg-Essen, Essen; Coordinating Centre for Clinical Trials (KKS), Philipps University, Marburg; and **Department of Pathology, Philipps University Marburg, Germany Background & Aims: The prognostic role of plasma chromogranin A in patients with neuroendocrine tumors is unclear. We investigated the role of chromogranin A in predicting survival and hypothesized that chromogranin A mirrors tumor burden and that a rapid increase after a phase of stable plasma chromogranin A levels might predict exploding tumor growth. Methods: Three hundred forty-four patients with metastatic, well-differentiated neuroendocrine tumors were included. A subsample of 102 patients was investigated to correlate radiologically classified tumor burden with plasma chromogranin A. Hepatic tumor burden (0%, 0% 25%, 25% 50%, >50%) was assessed from computed tomography/magnetic resonance imaging scans. Follow-up information until death was generated in regular intervals. Results: Plasma chromogranin A levels (U/L) vary between tumor entities (Kruskal Wallis, P <.001) and were associated with survival time (hazard ratio [hours], 2.14 per one unit in the log 10 CgA level scale; 95% confidence interval [CI], ; P <.001). Chromogranin A levels correlated with hepatic tumor burden (Spearman P.57; 95% CI, ; P <.001). Additional extrahepatic tumor load did not relevantly affect plasma chromogranin A. A sudden increase observed in individual patients was paralleled by rapid tumor progress and short survival. Conclusions: Increased plasma chromogranin A in patients with metastatic neuroendocrine tumors is predictive for shorter survival. There was a modest correlation between chromogranin A levels and hepatic tumor burden. We hypothesized further that a sudden increase in individual chromogranin A levels indicates unfavorable outcome. Growth of well-differentiated malignant neuroendocrine gastroenteropancreatic (GEP) tumors is mostly slow, 1 6 but despite well-differentiated histology, some tumors grow more rapidly, resulting in shorter survival. 5,6 In the clinical setting the prognosis of a patient with a newly diagnosed GEP tumor is usually unknown. Ki-67 estimation in biopsy material is currently the most reliable marker to predict tumor growth and often acts as surrogate marker of survival. 5,6 Serum tumor markers such as plasma chromogranin A (CgA), 7,8 neuropeptide K, 9 neurokinin A, 10 neuron-specific enolase, 11 the alpha-subunit of glycoprotein hormones, 12 and others have been tested to predict tumor growth. Currently, plasma CgA is widely used in following patients with metastatic tumors. CgA belongs to a family of at least 3 secretory proteins, Cg A, B, and C, with widespread distribution in the core vesicles of many normal and neoplastic neuroendocrine cells. 13,14 The highest levels of plasma CgA in the bloodstream have been found in patients with metastatic GEP tumors. 15 Furthermore, plasma CgA has been reported to be a sensitive tumor marker for monitoring patients with functionally active and inactive GEP tumors of pancreatic and extrapancreatic origin. 16,17 Recently, plasma neurokinin A has been shown to act as accurate and independent marker in assessing prognosis of midgut tumors and provided strong association with outcome. 10 We report here the experiences of a single institution with CgA as plasma marker in patients with metastatic well-differentiated GEP tumors. The purpose was 2-fold. First, we investigated the prognostic value of CgA plasma levels for survival. We expected that elevated CgA levels at diagnosis of metastatic disease are associated with shorter survival. Second, hepatic and extrahepatic tumor burden was assessed in a subsample of patients, and plasma CgA was determined simultaneously. In this cross-sectional evaluation we hypothesized that plasma CgA levels mirror tumor burden. Obtained information on survival of the subsample was analyzed exploratively. Materials and Methods Patients In 1988 a tumor register was initiated with data on clinical presentation and treatment, histopathology, and laboratory values of 641 patients with GEP tumors admitted in our institution. CgA levels reported in this study were used only if estimated in our laboratories by using an enzyme-linked immunosorbent assay that has not changed since January 1, Three hundred forty-four patients fulfilled this criterion and were used for the first part of the study investigating plasma CgA levels as a prognostic factor for survival (Table 1). All patients had computed tomography (CT) or magnetic resonance imaging (MRI) documented disease in the liver and/or within abdominal lymph nodes. Proof of a well-differentiated histology was performed by pathologic review (P.B.). A sub- Abbreviations used in this paper: CgA, chromogranin A; CI, confidence interval; CT, computed tomography; GEP, gastroenteropancreatic; MRI, magnetic resonance imaging by the AGA Institute /08/$34.00 doi: /j.cgh

2 July 2008 CHROMOGRANIN A IN NEUROENDOCRINE TUMORS 821 Table 1. Plasma CgA Levels and Survival: Clinical Characteristics of 344 Patients With Metastatic GEP Tumors for Which Plasma CgA Levels Were Available After 1995 Tumor group n (observed deaths) Female/male Median age (y) a Mean time diagnosis-cga measure (days) a Nonfunctioning pancreatic 66 (37) 33/ (47.6, 63.5) 139 (36, 486) Functioning pancreatic b 22 (8) 13/ (46.4, 55.5) 193 (17, 913) Nonfunctioning midgut 58 (16) 27/ (44.0, 67.0) 139 (32, 646) Functioning midgut (cs c ) 139 (73) 68/ (53.4, 67.5) 229 (51, 787) Hindgut 9 (3) 3/ (42.3, 68.9) 370 (112, 2301) Other d 50 (20) 30/ (46.6, 61.7) 234 (45, 997) Total 344 (157) 174/ (49.7, 65.6) 162 (41, 760) NOTE. Because of their heterogeneity or the small numbers, the groups in italics were not included in the Cox regression analysis with tumor entity as predictor. a 25% and 75% quartiles. b Comprises 6 insulinomas, 8 gastrinomas, 4 VIPomas, 2 glucagonomas, 1 PPoma, and 1 other. c Carcinoid syndrome. d Comprises the following locations of the primary: 11 bronchus, 8 duodenum, 1 jejunum, 3 stomach, and 27 of unknown origin. sample of 112 patients registered between July and November 2004 was included into the second part of the study investigating the correlation of plasma CgA and tumor burden. Patients were included if they had metastatic disease of a well-differentiated GEP tumor. Exclusion criteria included undifferentiated histology, chemoembolization, chemotherapy, and/or operation during the preceding 4 weeks, a time difference between imaging (CT or MRI) and plasma CgA determination of more than 4 weeks, as well as no metastatic tumor burden at the time of admission, resulting in a final subsample of 102 after exclusion of 10 patients. Finally, we determined for descriptive purposes plasma CgA levels in an additional sample of 95 patients admitted to the Endocrine Unit of our hospital. This cohort was free of an endocrine disorder including a nonfunctioning GEP tumor as shown by blood screening and imaging procedures. Chromogranin A Enzyme-Linked Immunoassay CgA from ethylenediaminetetraacetic acid plasma samples was determined with an ELISA Kit of DakoCytomation (Glostrup, Denmark) according to the information provided by the user manual. In the assay a rabbit polyclonal antibody directed against a 23 kd C-terminal fragment of human CgA was used, and the CgA standards of the Kit were calibrated against the C-terminal fragment. Median and normal range as indicated by the producer are 10 U/L (range, 2 18). Survival Follow-up information until death was generated once yearly in all registered patients by using a structural questionnaire sent to the responsible physician or by telephone calls. In the first part, survival time was calculated as the difference between the date of the first plasma CgA measurement after detection of metastatic disease and the date of death independent of cause or the last observation date in case of censoring. Furthermore, sensitivity analyses were performed (data not shown), with survival times calculated between the date of the first diagnosis of a metastatic GEP tumor and the respective last observation. The results were similar to those presented. To address the second question, updated survival information was obtained for the subsample, and survival time was calculated from the date of plasma CgA assessment at the time of admission. Evaluation of Tumor Load All patients of the first part of the study had measurable disease in the liver or within abdominal lymph nodes. CT or MRI scans of the abdomen and chest performed in all patients included into the second part of the study were evaluated for localization and size of the primary, extent of lymph node and hepatic metastases, for coexisting peritoneal carcinosis, for metastases within the lung, bone, and elsewhere. With a semiquantitative 3-dimensional approach, one of the authors (K.J.K.) judged the hepatic tumor load from 4 6 most relevant CT/MRI scans. He was blinded against CgA levels determined by an independent investigator (P.H.K.). Hepatic tumor burden was categorized as 0%, 0% but 25%, 25% but 50%, and 50% (abbreviated 0%, 0% 25%, 25% 50%, 50%). In addition, extrahepatic metastatic load in lymph nodes, lung, bone, and elsewhere was recorded (yes/no) but was not analyzed quantitatively (Table 2). Statistics Univariate, multivariate, and stratified Cox regression analyses were used to evaluate the association between survival time as defined above and both the predictor (log 10 of plasma CgA in U/L) and other covariables (gender, age, tumor group). The distribution of the variable plasma CgA levels is skewed toward higher values. The order-preserving log-transformation has been performed to draw near to a normally distributed variable both for analyses in regression models and for presentation of results. Model diagnostics comprised graphic and formal judgments of a potential violation of the proportional hazards assumption, an investigation of influential observations, as well as a check for nonlinearity (martingale residuals). The survival plots were based on the Kaplan-Meier method, with a plasma CgA level 200 U/L as cutoff. To obtain approximately equal numbers of patients within each group, the cutoff was chosen as rounded value of the estimated median (210 U/L in all 344 patients). In a stratified analysis the median plasma CgA levels per stratum were used as cutoff.

3 822 ARNOLD ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 7 Table 1. Continued Median survival (y) (95% CI) 5 Year survival (%) Total 200 U/L CgA levels 200 U/L CgA levels Total 200 U/L CgA levels 200 U/L CgA levels 2.9 ( ) 6.5 ( ) 2.0 ( ) ( ) ( ) 7.0 ( ) 2.1 ( ) ( ) 7.0 ( ) 2.4 ( ) For the second question, a sample size calculation was done before sampling. A sample size of 100 subjects was estimated to result in a power of 0.8 for rejecting the null hypothesis of no association (2-sided test with.05), assuming that Pearson true correlation coefficient was r 0.3. For the analysis, Spearman correlation coefficient was used to describe the relationship between hepatic tumor load as an ordinal measure (0%, 0% 25%, 25% 50%, 50%) and log 10 (CgA). Similar to the first research question, the relationship between tumor load, log 10 (CgA), and survival was explored. All P values reported are 2-sided, and a level.05 was considered to indicate statistical significance for each of the 2 primary analyses. SAS (version 8.2; SAS Institute Inc, Cary, NC), SPSS (version 12; SPSS Inc, Chicago, IL), StatXact (version 6.2.0; Cytel, Inc, Cambridge, MA), and R (version 2.4.1) were used for statistical analyses. Ethics The study was conducted in accordance to the ethical principles of the Declaration of Helsinki and was approved by the local ethical committee. Role of Funding Source Novartis Pharma provided funding but did not participate in the collection, analysis, and interpretation of the data. Results Characteristics of Enrollees Tables 1 and 2 provide descriptive information on patient characteristics of both parts of the study. For 6 patients the primary was unknown (Table 2). However, the exclusion of a pancreatic or bronchial origin according to CT, MRI, and Table 2. Plasma CgA Levels and Tumor Load: Clinical Characteristics of 102 Patients With Metastatic Well-Differentiated Neuroendocrine Tumors of Different Origin Tumor n Females/males Median age (y) (quartiles) Patients with hepatic/ extrahepatic metastases Median duration of disease (mo) (range) Foregut Bronchus Nonfunctioning 3 2/ / Functioning (cs) 1 0/ / Pancreas Nonfunctioning 22 9/ (42.94, 62.06) 20/ (23.5, 75.6) Functioning Insulinoma 1 1/ /0 1.5 Gastrinoma 3 2/ / VIPoma 2 2/ / Glucagonoma 1 0/ / Duodenum Nonfunctioning 1 1/ Functioning 0/1 Gastrinoma 2 0/ / Stomach Nonfunctioning 1 0/ / Midgut Nonfunctioning 34 16/ (44.13, 62.74) 28/ (19.1, 106.3) Functioning (cs) 23 15/ (46.89, 60.16) 19/ (13.0, 124.4) Hindgut 2 1/ / Unknown Nonfunctioning 6 4/ (45.45, 55.04) 4/ (3.7, 138.8) primary Total / (46.30, 62.06) 85/ (18.4, 85.0) cs, carcinoid syndrome.

4 July 2008 CHROMOGRANIN A IN NEUROENDOCRINE TUMORS 823 Table 3. Univariate and Multivariate Cox Analyses Univariate Multivariate (stratified by tumor group) Predictor and covariables Hazard ratio (95% CI) P value Hazard ratio (95% CI) P value log 10 (CgA) (U/L) 2.04 ( ) ( ).001 Gender Female 1.91 Male 1.02 ( ) Age at CgA measure (10 y) 1.21 ( ) ( ).93 Tumor group Nonfunctioning pancreatic 1 Nonfunctioning midgut 0.49 ( ).02 Functioning midgut 1.01 ( ).98 (Carcinoid syndrome) NOTE. Hazard ratio point estimates and 95% CIs and P values (2-sided) from the Wald tests are reported. Octreoscan suggested that these tumors originated from the midgut. Both subsamples listed in Tables 1 and 2 consisted primarily of midgut and nonfunctioning pancreatic tumors. Plasma Chromogranin A as a Prognostic Marker for Survival As shown in Tables 1 and 3, elevated plasma CgA levels were associated with shorter median and 5-year survival times. In the univariate Cox regression analysis, the log 10 -transformed plasma CgA level was a prognostic factor for survival. Figure 1 illustrates this finding, with a CgA cutoff of 200 U/L, which is the rounded value of the estimated median in all 344 patients. Although there was no evidence for a prognostic value of the covariate gender, there was some evidence for a prognostic value of the covariates age at first CgA measurement and tumor group. Thus, to underline the findings concerning plasma CgA, we adjusted these latter covariates in a multivariate analysis with stratification for tumor group. In this stratification, patients hindgut tumors have been excluded because 8 of 9 patients displayed CgA levels in the range of the control group. This analysis did not relevantly alter the results of the univariate comparison. Elevation of plasma CgA levels per se and not a tumor subgroup within GEP tumors was of prognostic significance (Table 3). The plasma CgA values of all 344 patients as well as those of the 95 controls are shown in Figure 2. A descriptive and explorative comparison of plasma CgA levels revealed that 5 of 6 pair-wise comparisons with the control patients indicated higher CgA levels in patients with a GEP tumor (all P.001 with the Mann-Whitney test). With the exception of the group of 9 patients with metastatic hindgut tumors (median CgA level, U/L; range, ), the CgA levels of the other groups were clearly elevated compared with controls (median, U/L; range, ), with the Figure 1. Kaplan Meier survival plots comparing 168 patients with plasma CgA levels 200 U/L against 176 patients with plasma CgA levels 200 U/L. Vertical line marks indicate censoring. Numbers underneath the plots indicate the size of the respective group at risk. Figure 2. Box plots of log 10 -transformed plasma CgA levels (log 10 (CgA), CgA in U/L) for the investigated tumor groups of 344 patients with metastatic, well-differentiated neuroendocrine tumors (see Table 1) and 95 control patients.

5 824 ARNOLD ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 7 highest levels observed for patients with carcinoid syndrome. For patients with metastatic hindgut tumors, the test failed to show a difference. However, one should note that only 9 patients (with normal CgA plasma levels in 8 patients) were included in this tumor group. Plasma CgA as a Marker for Tumor Load and Survival Figure 3A displays the correlation of plasma CgA levels with the extent of hepatic tumor burden. Spearman coefficient of correlation was 0.57 (95% confidence interval [CI], ; P.001). In patients with no hepatic tumor spread, eg, after curative resection of a malignant neuroendocrine tumor, plasma CgA (median, 25 U/L; range, U/L) was lower than in patients with a hepatic tumor load not exceeding 25% (median, 48 U/L; range, U/L). Plasma CgA levels increased further in patients with hepatic tumor involvement between 25% and 50% (median, 704 U/L; range, 13 14,710 U/L) and were found to be highest in patients with a hepatic tumor load 50% (median, 1009 U/L; range, U/L). Explorative analysis revealed that patients with functionally active tumors had higher plasma CgA levels compared with patients with nonfunctioning tumors (Figure 3B). Additional extrahepatic lymph node metastases in the presence of liver involvement did not relevantly alter, ie, further increase, plasma CgA levels (data not shown). We explored the relationship between survival, hepatic tumor load, and plasma CgA by conducting Cox regression analyses, with stratification according to the tumor load of the liver (0%, 0% 25%, 25% 50%, 50%) (Figure 4). In each stratum of tumor load, we estimated the median plasma CgA level and used this value as cutoff in the survival analysis. Univariate analysis revealed that patients prognosis regarding survival decreased with increasing hepatic tumor load, with the shortest survival in patients with a hepatic tumor load 50% (hazard ratio 1.43 for shifting to the next higher category (95% CI, ; P.09). In accordance with the results described above, plasma CgA levels were correlated with survival (hazard ratio 3.71 per one unit in the log 10 of CgA level scale; 95% CI, ; P.001). Joint modeling of both factors, however, revealed that hepatic tumor load did not independently determine prognosis of survival if compared with plasma CgA levels estimated simultaneously (CgA, 4.83; 95% CI, ; P.001; tumor load, 0.69; 95% CI, ; P.15). Further to the observation that elevated plasma CgA levels have a stronger influence on survival compared with hepatic tumor load if both parameters were evaluated simultaneously, we obtained additional evidence for the assumption that a sudden and rapid increase of plasma CgA in patients with metastatic disease serves as indicator of exaggerated tumor progression. This is illustrated in Table 4, with 4 patients each with a highly differentiated GEP tumor. was chosen as the rounded value of 210 U/L, the median of plasma CgA for the whole group of patients. It is unclear from published data which cutoff discriminates best the patients with a more favorable from those with a less favorable prognosis. In a multivariate analysis, Janson et al 17 found in 71 tumor patients that a plasma CgA level 5000 g/l was an independent predictor of overall survival. From the results obtained in the second part of our study, we concluded that there is no absolute CgA level discriminating tumor patients with a more favorable from those with an unfavorable prognosis. Even in patients with high hepatic tumor burden, those with CgA plasma levels below the median of plasma CgA of the respective Discussion This study demonstrated in accordance with previous reports that plasma CgA is a valuable marker for malignant GEP tumors of different origins, with the highest CgA plasma levels in patients with carcinoid syndrome. 8,10,15 20 As shown in the first part of this study, survival was reduced if plasma CgA levels were 200 U/L at the time of diagnosis. The cutoff value Figure 3. (A) Correlation of plasma CgA levels (log 10 [U/L]) and hepatic tumor burden for 17 patients without hepatic but with extrahepatic metastases, 58 patients with liver tumor burden 0% and 25%, 15 patients with liver tumor burden 25% and 50%, and 12 patients with liver tumor burden 50%. Correlation of plasma CgA levels (log 10 (CgA), CgA in U/L) and hepatic tumor burden, with additional stratification by functioning/nonfunctioning tumor entity (B).

6 July 2008 CHROMOGRANIN A IN NEUROENDOCRINE TUMORS 825 Figure 4. Kaplan-Meier survival plots stratified by hepatic tumor burden (same patients as in Figure 3A). Within each group a median split for plasma CgA was performed. group had a more favorable (survival) prognosis compared with patients with higher CgA levels (Figure 4). We could further show that plasma CgA levels mirror hepatic tumor load, with the highest plasma levels found in patients with a hepatic tumor burden exceeding 50%. Additional lymph node involvement or metastases elsewhere did not relevantly alter plasma CgA levels, presumably as a result of the preponderance of the hepatic tumor involvement. Janson et al 17 reported that patients with 5 liver metastases had higher CgA plasma levels and a shorter survival compared with patients with 5 liver metastases. We extended these observations by determining hepatic tumor load by using a semiquantitative 3-dimensional approach. Computer-based estimation of hepatic tumor volume is currently still under investigation and far from being validated for the clinical setting. 25 Therefore, one of the authors (K.J.K.) judged the tumor load according to his personal experience in the field of CT/MRI imaging of GEP tumors. 6 Counting the number of liver metastases as done by others 17 does not recognize the size of an individual lesion, and 5 small metastases correspond to a much lower tumor load than 5 large metastases. Taken into account the possibility of measurement errors, we could with our procedure demonstrate that CgA levels correlate well with hepatic tumor load. Our findings are also in line with experimental findings in nude mice with xenografted human ileal tumors, 21 showing that CgA plasma levels closely correlate with tumor burden. In contrast, Bashir et al 26 did not find a significant relationship between serum CgA levels and tumor extent in gastrinoma patients, which might be due to the additional secretion of CgA from gastric enterochro- Table 4. Course of Plasma CgA Levels in 4 Patients With Metastatic, Well-Differentiated Neuroendocrine Tumors Patient Plasma CgA (U/mL) Hepatic tumor burden (%) Duration of disease (mo) 1 a Died 96 2 b Died c , Died 5 4 d Died 74 a Midgut tumor initially with regional lymph node involvement only. b Carcinoid syndrome. c Nonfunctioning tumor with unknown primary. d Bronchial neuroendocrine tumor.

7 826 ARNOLD ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 7 maffin-like cells stimulated by gastrin from the tumor. Accordingly, serum gastrin correlates strongly with serum CgA levels in gastrinoma patients. 26 We found in 3 heavily metastatic gastrinomas included in this study still a correlation between hepatic tumor load and CgA plama levels. Finally, we explored the relationship between hepatic tumor load, plasma CgA levels, and survival. We could demonstrate that tumor load and plasma CgA levels are both independent prognostic factors for survival. However, a multivariate analysis indicated a higher influence of plasma CgA on survival if assessed simultaneously with hepatic tumor burden. Our 4 case reports listed in Table 4 suggested, in addition, that a sudden increase of plasma CgA exceeding 1000 U/L indicates rapid tumor growth and an unfavorable outcome. This occurred in 3 of 4 patients with well-differentiated GEP tumors after longlasting phases of stable tumor disease: 80 months in patient 1, 123 months in patient 2, and 58 months in patient 4. Rapid tumor growth resulted in hepatic tumor involvement of 50%. We do not know whether tumor growth in patient 3 was rapid from the beginning of the disease. More likely, a longer-lasting phase of slow tumor progression associated with no or only minor symptoms might have preceded the actual phase of exaggerated tumor growth in this patient. Thus, a tumor burden 50% is associated with a fatal prognosis for survival if tumor burden developed very fast and is mirrored by a rapid increase of plasma CgA levels. Vice versa, high CgA levels in patients with high hepatic tumor load do not necessarily indicate approaching fatal prognosis if tumor progression and elevated CgA levels developed slowly. In contrast to others, 17 plasma CgA levels in our patients with metastatic hindgut tumors were elevated only in 1 of 9 patients. This was obviously not the consequence of the assay used. By comparing 3 commercial assays including the assay used by us, a strong positive linear correlation was found between serum and plasma CgA levels, with variations in the absolute levels The sensitivity of the kits in confirming elevated plasma CgA levels varied between 67% and 93% and the specificity between 85% and 96%. 23 The lack of international standards for CgA and comparable antibodies impaired the comparison of CgA levels on the basis of different kits. Despite these limitations, available assays are, in principle, able to answer important clinical questions as to the usefulness to act as tumor marker. In conclusion, our data supported the view that with the exception of hindgut tumors, elevated plasma CgA levels are of prognostic relevance for survival in patients with malignant well-differentiated GEP tumors. CgA plasma levels mirror hepatic tumor burden, with the highest levels in patients with the highest hepatic tumor burden. We further hypothesized that a rapid increase of plasma CgA with values exceeding 1000 U/L after a longer-lasting phase of stable tumor disease with lower CgA plasma levels might act as indicator of rapid tumor progression and unfavorable outcome. References 1. Arnold R, Simon B, Wied M. Treatment of neuroendocrine GEP tumours with somatostatin analogues. Digestion 2000;62(Suppl 1): Godwin JD. Carcinoid tumors: an analysis of 2837 cases. Cancer 1975;36: Williams ED, Sandler M. The classification of carcinoid tumors. Lancet 1963;75: Modlin IM, Lye KD, Kidd M. A 5-decade analysis of 13,715 carcinoid tumors. Cancer 2003;97: Rindi G, Azzoni C, La Rosa S, et al. ECL cell tumor and poorly differentiated endocrine carcinoma of the stomach: prognostic evaluation by pathological analysis. Gastroenterology 1999; 116: Arnold R, Rinke A, Klose K-J, et al. Octreotide versus octreotide plus interferon-alpha in endocrine gastroenteropancreatic tumors: a randomized trial. Clin Gastroenterol Hepatol 2005;3: Hsiao RJ, Seeger RC, Yu AL, et al. Chromogranin A in children with neuroblastoma. Serum concentration parallels diease stage and predicts survival. J Clin Invest 1990;85: Nobels FR, Kwekkeboom DJ, Coopmans W, et al. Chromogranin A as serum marker for neuroendocrine neoplasia: comparison with neuron-specific enolase and the alpha-subunit of glycoprotein hormones. J Clin Endocrinol Metab 1997;82: Norheim I, Theodorsson-Norheim E, Brodin E. et al. Tachykinins in carcinoid tumours: their use as a tumor marker and possible role in the carcinoid flush. J Clin Endocrinol Metab 1986;63: Turner GB, Johnston BT, McCance DR. et al. Circulating markers of prognosis and response to treatment in patients with midgut carcinoid tumours. Gut 2006; 55: Prinz RA, Bermes EW, Kimmel JR. et al. Serum markers for pancreatic islet cell and intestinal carcinoid tumors: a comparison of neuron specific enolase, -human chorionic gonadotropin and pancreatic polypeptide. Surgery 1983;94: Gro mann M, Trautmann ME, Poertl S, et al. Alpha-subunit and human chorionic gonadotropin- immunoreactivity in patients with malignant endocrine gastro-entero-pancreatic tumors. Eur J Clin Invest 1994;24: Taupenot L, Harper KL, O=Connor DT. The chromogranin-secretogranin family. N Engl J Med 2003;348: O=Connor DT, Frigon RP. Chromogranin A, the major catecholamine storage vesicle soluble protein: multiple size forms, subcellular storage, and regional distribution in chromaffin and nervous tissue elucidated by radioimmunoassay. J Biol Chem 1984; 259: Öberg K, Tiensuu Janson E, Erikson B. Tumor markers in neuroendocrine tumours. Ital J Gastroenterol hepatol 1993;3 (Suppl 2):S160 S Goebel SU, Serrano J, Yu F, et al. Prospective study of the value of serum chromogranin A or serum gastrin levels in the assessment of the presence, extent, or growth of gastrinomas. Cancer 1999;85: Janson ET, Holmberg L, Stridsberg M, et al. Carcinoid tumors: analysis of prognostic factors and survival in 301 patients. Ann Oncol 1997;8: Öberg K, Tiensuu Janson E, Erikkson B. Tumour markers in neuroendocrine tumours. Ital J Gastroenterol Hepatol 1999;3: S160 S Stivanello M, Berruti A, Torta M, et al. Circulating chromogranin A in the assessment of patients with neuroendocrine tumours: a single institution experience. Ann Oncol 2001;12(Suppl 2):S73 S Nikou GC, Lygidakis NJ, Toubanakis C, et al. Current diagnosis and treatment of gastrointestinal carcinoids in a series of 101 patients: the significance of serum chromogranin A, somatostatin receptor scintigraphy and somatostatin analogues. Hepatogastroenterology 2005;52: Kölby L, Bernhardt P, Sward C, et al. Chromogranin A as a determinant of midgut carcinoid tumour volume. Regul Pept 2004;120: Ardill JES, Erikkson B. The importance of the measurement of circulating markers in patients with neuroendocrine tumours of the pancreas and gut. Endocr Relat Cancer 2003;1:

8 July 2008 CHROMOGRANIN A IN NEUROENDOCRINE TUMORS Stridsberg M, Erikkson B, Öberg K, et al. A comparison between three commercial kits for chromogranin A measurements. J Endocrinol 2003;177: Woltering EA, Hilton RS, Zolfoghary CM, et al. Validation of serum versus plasma measurements of chromogranin A levels in patients with carcinoid tumors: lack of correlation between absolute chromogranin A levels and symptom frequency. Pancreas 2006; 33: Gluer CC, Barkmann R, Hahn HK, et al. Parametric biomedical imaging: what defines the quality of quantitative radiological approaches? RoFo 2006;178: Bashir S, Gibril F, Ojeaburu JV, et al. Prospective study of the ability of histamine, serotonin or serum chromogranin A levels to identify gastric carcinoids in patient with gastrinomas. Aliment Pharmacol Ther 2002;16: Address requests for reprints to: Prof Dr med Rudolf Arnold, Wittelsbacherstr 6, München, Germany. arnoldr@mailer.unimarburg.de;; fax: Supported by grants from Novartis, Nürnberg, Germany.