Clinical Chemistry 53:3 438 – 446 (2007)

Cancer Diagnostics

Processing-Independent Quantitation of Chromogranin A in Plasma from Patients with Neuroendocrine Tumors and Small-Cell Lung Carcinomas Tine Børglum,1 Jens F. Rehfeld,1 Lars B. Drivsholm,2 and Linda Hilsted1* Background: Most neuroendocrine tumors express chromogranin A (CgA). The posttranslational processing of neuroendocrine proteins such as CgA is often specific for the individual tumor. To cope with this variability and improve tumor diagnosis, we developed a processing-independent analysis (PIA) method to measure the total CgA product. Methods: For PIA, samples underwent trypsin treatment followed by measurement of CgA by the “CgA(3403)” assay, in which the antiserum binds an epitope starting at amino acid 340 of CgA and including amino acid residues located in the C-terminal direction. The diagnostic accuracy of the CgA PIA and 3 sequencespecific assays for CgA were evaluated on plasma samples from patients with neuroendocrine tumors and small-cell lung carcinomas. Furthermore, we investigated whether the CgA plasma concentrations correlated with the tumor burden. Results: Size-exclusion chromatography of plasma showed that CgA immunoreactivity mainly consisted of high–molecular-weight forms, indicating that neuroendocrine tumors may secrete large amounts of poorly processed CgA. Accordingly, trypsination of plasma from 54 patients with neuroendocrine tumors or smallcell lung carcinomas increased the CgA(3403) immunoreactivity up to 500-fold. Both the CgA(3403) assay and the PIA measured significantly higher plasma concentrations in patients with very extensive disease than in patients with less widespread disease. The diagnostic

sensitivity was 0.91 when using the CgA(3403) assay and 0.82 using the CgA PIA. Conclusion: The CgA(3403) assay and CgA PIA are both useful for diagnosis of neuroendocrine tumors and small-cell lung carcinomas and both assays correlate with tumor burden. © 2007 American Association for Clinical Chemistry

The human chromogranin A (CgA) protein, a member of the granin family, is an acidic 439 amino acid protein with a sequence containing several mono- and dibasic cleavage sites. CgA is normally found in endocrine cells and neurons and is a promising tumor marker because most neuroendocrine tumors, including silent tumors without secretion of known hormones, express and release CgA (1– 4 ). Processing of neuroendocrine tumor proteins is often heterogeneous and attenuated (5, 6 ); therefore an analytical principle has been developed to improve the diagnosis of neuroendocrine tumors. Processing-independentanalysis (PIA), which measures an epitope even when the epitope is hidden in precursors and processing intermediates, quantifies the entire translation product of a given protein system irrespective of the degree of posttranslational processing (7 ). PIA was first applied to progastrin and procholecystokinin (7–9 ), and recent reports illustrate its diagnostic advantages (5, 6, 10 –14 ). The purpose of this study was to develop a PIA for human CgA in plasma and to evaluate the diagnostic accuracy of measurement of the entire CgA translation product (total CgA) with the CgA PIA (index test) to differentiate between patients with gastrointestinal tumors, small-cell lung carcinoma (SCLC), or pheochromo-

1 University Department of Clinical Biochemistry, Rigshospitalet, Copenhagen, Denmark. 2 Department of Oncology, Storstrømmens Sygehus Næstved, Denmark. * Address correspondence to this author at: Department of Clinical Biochemistry KB, Rigshospitalet, 9 Blegdamsvej, DK-2100 Copenhagen Ø, Denmark. Fax 45-35-45-46-40; e-mail [email protected]. Received July 10, 2006; accepted December 29, 2006. Previously published online at DOI: 10.1373/clinchem.2006.076158

3 Nonstandard abbreviations: CgA, chromogranin A; PIA, processingindependent analysis; SCLC, small-cell lung carcinoma; Kds, coefficients of distribution; and CI, confidence interval.

438

Clinical Chemistry 53, No. 3, 2007

cytoma and healthy individuals. In addition we investigated whether the CgA plasma concentrations correlate with the tumor burden.

439

test, i.e., treated with trypsin, as described below, and subsequently analyzed with the CgA(3403) assay.

enzymatic treatment of plasma samples Materials and Methods nomenclature The specificity of the assays (antisera) is described by indicating the N- or C-terminal amino acid of the hapten as recognized by the antiserum. Thus, “CgA (3403) assay” indicates that the antiserum employed in this assay binds an epitope that starts at amino acid number 340 of CgA and includes amino acid residues located in the C-terminal direction.

immunoassays Index test: development of a PIA for human CgA. An amino acid sequence of 6 – 8 amino acid residues is selected for immunization. It should not undergo posttranslational modifications, and it must be located between suitable (tryptic) cleavage sites. The sample to be assayed is preincubated with trypsin. In addition to exposing the epitope, the endoproteolytic cleavage ensures that equalsized fragments with lengths similar to the assay calibrator are released so that proprotein and final products are quantitated with equimolar potency (5, 7, 9, 11 ). The CgA(340 –348) sequence neighbors suitable tryptic cleavage sites, i.e., trypsination will release the CgA fragment Leu340–Lys355. Trypsin treatment (described below) and subsequent quantification by the CgA(3403) assay thus provide an estimate of the total CgA mRNA translation product, i.e., total CgA. Furthermore the CgA (340 –348) sequence does not undergo posttranslational modifications. Sequence-specific CgA (340 –348) RIA. We previously developed an RIA specific for the N-terminus of the sequence CgA(340 –348) [CgA(3403)], adjacent to a dibasic site (Lys338Arg339) in the midportion of hCgA (using antibody 95058) (15 ). The interassay CVs of replicate samples for the CgA(3403) assay were 21% at 15 pmol/L and 18% at 62 pmol/L. Sequence-specific RIAs for CgA(1–9) and CgA(250 –301) amide. We previously developed an RIA specific for the N-terminus of human CgA [using antibody 94188 (CgA(13)] (15 ). The CgA(4301) assay (using antibody 871w-1) is specific for the amidated C-terminus of pancreastatin (16 ). The antiserum reacted fully with the synthetic peptides hCgA(273–301) amide and hCgA(250 –301)amide.

plasma samples Blood samples were collected as previously described (15 ). All plasma samples were analyzed with the CgA(13), CgA(4301 amide), and CgA(3403) assays. Furthermore, the samples were analyzed with the index

For trypsin cleavage, equal volumes of plasma and 0.1 mol/L sodium phosphate buffer (pH 7.5) containing 2.0 g trypsin/L were incubated at room temperature for 30 min. For enzymatic cleavage of column fractions a trypsin concentration of 0.2 g/L was used. The enzymatic reaction was terminated by boiling for 10 min. Parallel control experiments were performed by omission of trypsin from the sodium phosphate buffer.

size-exclusion chromatography of plasma samples Plasma samples from carcinoid tumor patients (150 –200 ␮L) were applied to a Superose 12 HR column. Chromatography was performed as described previously (15 ). The column was calibrated with synthetic hCgA(250 –301 amide) and an hCgA (340 –372) fragment purified from a midgut carcinoid tumor. The identity of the fragment was established by mass spectrometry analysis. The fractions were analyzed, before and after tryptic cleavage, with the CgA(3403) assay.

reference test: histological classification of neuroendocrine tumors and sclc For tumor patients histopathological analysis was performed as a routine test for diagnosis and classification. Biopsy material, surgical or coarse-needle biopsy specimens, was fixed in formalin and paraffin-embedded. The tumors were immunostained with a panel of antibodies to general neuroendocrine markers (e.g., CgA and synaptophysin) as well as specific hormones. The microscopy report included a general morphological description; immunohistochemical profile; mitotic rate; vascular, perineural, or lymphatic invasion; infiltration of surrounding tissues; lymph node status; and distant metastases (17–20 ). For 24, 14, 8, and 8 of the patients the reference test was performed at the Departments of Pathology, at Hillerød Hospital, Rigshospitalet, Herlev Hospital, and other Danish hospitals, respectively. The test was performed by trained technicians and read by medical specialists in pathology (n ⫽ 10). The readers of the reference test had access to the clinical information on the patients, but not to index test results.

evaluation of diagnostic accuracy of total cga (index test) Plasma samples from tumor patients. Included in this study were 27 patients with gastrointestinal neuroendocrine tumors, 6 patients with pheochromocytoma, and 21 patients with SCLC recruited consecutively upon referral for treatment of the tumors/carcinomas at the Departments of Surgical Gastroenterology and Urology, Rigshopitalet; the Department of Endocrinology, Herlev Hospital; and

440

Børglum et al.: Processing-Independent Assay for Chromogranin A

The study was approved by the local ethics committee and informed consent was obtained from all patients (KF01–393/95, KF01–352/96, KF01–290/97, KF01– 014/98).

Fig. 1. Flow diagram of patients and controls in study. Gate 1 inclusion criteria included an established histopathological tumor diagnosis (reference test), for further details see text. Chromogranin A was determined by CgA PIA (index test) and by 3 sequence-specific CgA assays.

the Department of Internal Medicine, Hillerød Hospital. Inclusion criteria for all patients were an established histopathological diagnosis (reference standard) made from biopsy material or tumor tissue. Exclusion criteria were initiated chemotherapy or concurrent diseases (Fig. 1, gate 1 criteria). For the patients with gastrointestinal neuroendocrine tumors the reference test was performed between May 1994 and October 1997, and the patients were recruited and the plasma samples for the index tests were collected at median (range) 2 (0.25–24) months after the reference test was performed. The patients did not receive chemotherapy in the time interval between the 2 tests, and for all these patients it was shown by computer tomography and/or ultrasonography that tumor tissue was present at the time of plasma sample collection. For the 6 pheochromocytoma patients the reference test was performed between August 1995 and September 1996. In these patients the plasma samples used for index tests were drawn shortly before surgery and reference testing. The reference tests on the 21 SCLC patients were made between May 1991 and November 1992, and the plasma samples for index tests were collected shortly after the reference test was performed. No treatment (chemotherapy or operation) took place between the 2 tests. The patient characteristics (age, sex, and severity of disease) are shown in Table 1.

Plasma samples from healthy controls. To establish reference intervals for the CgA(3403) and CgA(3403) posttryptic assays, we obtained plasma samples after 15-min rest from 88 healthy nonfasting control individuals. These controls were volunteers recruited from hospital staff. Only persons who reported being healthy (i.e., feeling well, without known diseases, and not taking any medications) were included (Fig. 1, gate 2 criteria). Control age and sex are shown in Table 1. The CgA response to a mixed meal was studied in 8 healthy volunteers (4 men and 4 women; age range, 23–50 years) after an overnight fast. Consecutive postprandial blood samples were drawn 5–180 min after ingestion of the meal and analyzed with the 4 CgA assays. For all patients and controls the index tests were performed at the Department of Clinical Biochemistry, Rigshospitalet, between January 1999 and July 2002. The index test was carried out by 2 trained technicians and read by the same technicians and a medical specialist in clinical biochemistry. The readers of the index test knew whether the plasma samples were from healthy persons or from tumor patients and also knew the result of the reference standard. Clinical information (symptoms, scan/operation results, and medication) on the patients was made available to the medical specialist after the index test had been performed.

statistical methods Kruskal–Wallis ANOVA, Mann–Whitney U-test (2tailed), and Wilcoxon rank sum test were performed with GraphPad Prism 4 for Windows; version 4.03, GraphPad Software Inc. For calculating reference values and performing the Kolmogorov–Smirnov test of normality we used SPSS for Windows v.12.0, SPSS Inc. The 0.95 confidence intervals (CIs) of the 2.5th and 97.5th percentiles were calculated according to Tietz et al. (21 ). We performed the exact binomial test to calculate 95% CIs for diagnostic sensitivity. Differences in diagnostic sensitivi-

Table 1. Characteristics of tumor patients and healthy controls. n Age, years Female, n (%) Severity of disease, n

a

Gastrointestinal tumors

Pheochromocytomas

SCLC

Healthy controls

27 60 (57–70)a 12 (44) Local disease,b 3 Ext. disease,d 16 Very ext. disease,e 8

6 42 (36–49)a 5 (83) Local disease,b 6

21 65 (57–68)a 9 (43) Local disease,c 12 Ext. disease, 9

88 41 (30–51)a 44 (50)

Data are given as median (IQR). Disease without known liver/lymph node metastases. c Disease confined to 1 hemithorax excluding proven malignant effusion and chest wall metastases. d Extensive disease: liver/lymph node metastases. e Arbitrarily defined as either several large metastases or multiple smaller metastases in the liver. b

441

Clinical Chemistry 53, No. 3, 2007

ties in subgroups of patients were tested for significance with Fishers exact test. The exact binomial test and Fishers exact test (2-tailed) were performed using R Version 2.3.1 (22 ).

Results tryptic digestion of plasma samples Trypsin treatment of plasma increased the median CgA(3403) immunoreactivity 9-fold in controls and 6- to 16-fold in patients with neuroendocrine tumors and SCLC (Table 2 and Fig. 2). However, a large variation in the increase of CgA(3403) immunoreactivity was found within the tumor groups, with up to 500-fold increases found in some patients.

tryptic digestion of chromatography fractions Size-exclusion chromatography profiles of plasma samples from patients with midgut carcinoid tumors showed that the majority of the CgA(3403) immunoreactivity eluted from a Sepharose HR 12 column with coefficients of distribution (Kds) of 0.6 – 0.7. After trypsin treatment of the chromatography fractions, the majority of the posttryptic CgA immunoreactivity eluted in broad peaks (Kd 0.15– 0.45, maximum 0.3) that comprised several high– molecular-weight forms, and only a minor fraction eluted

corresponding to the peaks with Kds of 0.6 – 0.7 (see Fig. 1 in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/ content/vol53/issue3). These results indicated that the plasma samples in which trypsination markedly increased CgA(3403) immunoreactivity contained high amounts of unprocessed or poorly processed CgA.

quantification of cga in plasma In agreement with a previous study (23 ), meal ingestion in the controls did not lead to significant changes in plasma CgA(13), CgA(4301), CgA(3403), or posttryptic CgA(3403) (Wilcoxon, P ⬎0.05, data not shown). Consequently, reference intervals were established in a nonfasting population. The distributions of ln-transformed CgA(3403) and ln-transformed CgA(3403) posttryptic data were gaussian (Kolmogorov–Smirnov test of normality, P ⫽ 0.2 for both distributions). Reference intervals (central 95%) were 29 –132 mol/L (95% CIs: 27–31 and 122–143) for CgA(3403) immunoreactivity and 250 –1100 pmol/L (95% CIs: 230 –270 and 1020 –1195 pmol/L) for CgA(3403) posttryptic immunoreactivity. The reference interval for the CgA(13) assay was 288 – 696 pmol/L, and for the CgA(4301) assay was 3–15 pmol/L (15 ).

Table 2. Concentrations measured in plasma samples from tumor patients with local and extensive disease. Tumor

GI NE tumors All (n ⫽ 27) 1. Local (n ⫽ 3) 2. Extensive (n ⫽ 16) 3. Very extensive (n ⫽ 8) Pheochromocytoma 1. Local (n ⫽ 6) SCLC All (n ⫽ 21) 1. Local (n ⫽ 12) 2. Extensive (n ⫽ 9) Healthy volunteers (n ⫽ 88)

CgA(3403) pmol/L

CgA(3403)ⴙtrypsin pmol/L

CgA(3403) ⴙ trypsin/CgA(3403)

720 (287–2947) 140a (140–840)d 335a (222–2012) 4450a (2235–11 650)

4200 (1400–25 200) 900b (890–4200)d 2255b (1390–10 860) 102 050b (20 767–276 500)

7 (4–15) 6c (5–6)d 7c (4–8) 16c (12–21)

380 (262–650)

2750 (1110–10 850)

7 (3–24)

300 (130–440) 150e (120–325) 480e (245–825) 60 (48–78)

2215 (1117–8535) 1820f (1037–2278) 7150f (2840–10 850) 510 (420–660)

10 (8–18) 9 (7–18) 11 (9–16) 9 (6–11)

Results are given as median (IQR) exceptd where results are median (range) due to n ⫽ 3. Gastrointestinal neuroendocrine (GI NE) tumors: 3a Significantly different from: 2a(P ⫽ 0.009), 1a and 2a (P ⫽ 0.005), 1a(P ⫽ 0.02). 3b Significantly different from: 2b(P ⫽ 0.001), 1b and 2b (P ⫽ 0.001), 1b(P ⫽ 0.01). 3c Significantly different from: 2c(P ⫽ 0.001), 1c and 2c (P ⫽ 0.001), 1c(P ⫽ 0.01). SCLC: 2e Significantly different from 1e (P ⫽ 0.01). 2f Borderline significantly different from 1f (P ⫽ 0.06)

442

Børglum et al.: Processing-Independent Assay for Chromogranin A

evaluation of diagnostic accuracy

Fig. 2. CgA concentrations, measured by 4 CgA assays, in plasma samples from patients with carcinoid tumors, SCLC, pheochromocytoma, or other neuroendocrine tumors and in healthy controls. Horizontal lines indicate upper limit of the corresponding reference interval: CgA(13) 696 pmol/L, CgA(4301) 15 pmol/L, CgA(3403) 132 pmol/L, and CgA(3403) ⫹ trypsin 1100 pmol/L. Plasma samples were incubated with trypsin before CgA(3403) assay, as described in the text.

The concentrations measured with the 4 CgA assays in the 54 plasma samples from the tumor patients and controls are shown in Table 3 and Fig. 2. In defining the diagnostic sensitivity (the ratio of true positive to true positive plus false negative) true-positive vs true-negative designations were applied to patients with a neuroendocrine tumor diagnosis established by histopathological examination, depending on whether or not they had increased plasma CgA concentrations compared to controls. The upper limits of the reference intervals (central 95%) were used as cutoffs. The diagnostic sensitivities of tumor patients are shown for the 4 CgA assays in Table 4A. In all groups the highest diagnostic sensitivities were obtained with the CgA(3403) assay and CgA PIA. The diagnostic sensitivity obtained with the CgA(3403) assay was 1.00 for patients with local, extensive, and very extensive gastrointestinal neuroendocrine tumors. For all other assays the diagnostic sensitivities increased with increasing tumor burden (Table 4B). However, significant differences between the diagnostic sensitivities of these groups were found only for the CgA(13) assay and the CgA PIA. Furthermore, significant differences between the diagnostic sensitivities obtained for limited and extensive SCLC were found only for the CgA(3403) assay (Fishers exact test, Table 4B). In the patients with gastrointestinal tumors a higher tumor burden was generally accompanied by higher concentrations of CgA(3403) and total CgA immunoreactivity, as well as a lower degree of CgA processing. Kruskal–Wallis ANOVA applied on the CgA(3403), CgA(3403) ⫹ trypsin, and ratio [CgA(3403) ⫹ trypsin/ CgA(3403)] results from the 3 groups of gastrointestinal tumor patients with limited (group 1), extensive (group 2), and very extensive (group 3) disease showed significant differences between the groups, P ⫽ 0.009 [CgA(3403)], P ⫽ 0.002 [CgA(3403) ⫹ trypsin], and P ⫽ 0.002 [CgA(3403) ⫹ trypsin/CgA(3403)], respectively. Posthoc Mann–Whitney U-test demonstrated significant differences between group 3 vs 2, 3 vs 1 and 2, and 3 vs 1 for the 3 sets of data [CgA(3403), CgA(3403) ⫹ trypsin, and CgA(3403) ⫹ trypsin/CgA(3403)], whereas group 1 did not differ significantly from group 2 or from groups 2 and 3 (Table 2). For the patients with SCLC, the concentrations measured with the CgA(3403) assay in plasma from patients with extensive disease (group 2) were significantly higher than the concentrations measured in patients with local disease (group 1) (Mann–Whitney U-test, Table 2). The corresponding concentrations measured with the CgA PIA were borderline significantly higher (P ⫽ 0.06, Table 2).

Discussion Our results demonstrate that both the CgA(3403) and posttryptic CgA(3403) assays have a high diagnostic sensitivity in patients with neuroendocrine tumors or

443

Clinical Chemistry 53, No. 3, 2007

Table 3. Concentrations measured with the index test (CgA PIA) and 3 sequence-specific CgA assays in plasma samples from tumor patients with established histopathological diagnosis (reference standard). Reference standard (histopathology)

Index test (CgA PIA) Sequence-specific CgA assays

CgA(3403) ⫹ trypsinb CgA(13) CgA(4301) CgA(3403)

Carcinoid (n ⴝ 22)

Other GI NE tumorsa (n ⴝ 5)

3730c (1390–23 200)d 698 (383–2825) 34 (9–313) 620 (289–3240)

9050 (1370–102 050) 780 (523–7875) 12 (8–284) 1170 (140–4905)

Pheochromocytoma (n ⴝ 6)

2750 (1110–10 850) 840 (590–1503) 15 (5–18) 380 (252–650)

SCLC (n ⴝ 21)

2215 (1117–8535) 610 (503–1025) 16 (5–34) 300 (130–440)

a

GI NE, gastrointestinal neuroendocrine. Plasma samples were incubated with trypsin prior to CgA(3403) assay, as described in the text. c All results are given in pmol/L. d Results are given as median (IQR). b

SCLC. In agreement with these results, high concentrations of unprocessed CgA were found in plasma from these patients, as corroborated by size-exclusion chromatography. These results are in accordance with reports that PIA analyses have increased the diagnostic accuracy for progastrin (6, 10 ), probrain natriuretic peptide, and rat epidermal growth factor (11, 24 ). Furthermore, a PIA for procholecystokinin has recently been found to be a valuable tool for the diagnosis and monitoring of the efficiency of chemotherapy for patients with Ewing sarcomas, and the PIA proved superior to a conventional cholecystokinin RIA (14 ). An application of PIA for human CgA has not previously been published. The observation that the increase in CgA(3403) immunoreactivity after trypsination of plasma samples was significantly higher in plasma from patients with very extensive gastrointestinal neuroendocrine tumors (group 3) than in plasma from the patients with extensive (group 2) and local disease (group 1) (Table 2) is in agreement with earlier reports (5, 8 ) in which a lower degree of processing of progastrin was found in plasma from gastrinoma patients with metastases. A low degree of progastrin processing in gastrinomas has been previously suggested to predict a malignant clinical course (8 ), whether this could be the case for CgA from gastrointestinal tumors requires further study. Furthermore, our finding that the concentrations measured with the CgA(3403) assay and the PIA increased with the degree of tumor/carcinoma dissemination is in agreement with a previous report (25 ) that CgA concentrations were significantly higher in plasma from midgut carcinoid tumor patients with ⱖ5 liver metastases than in those with ⬍5 liver metastases. Plasma CgA has also been reported to be significantly higher in metastatic gastroenteropancreatic endocrine tumors than in nonmetastatic tumors (26 ), and higher in patients with extensive SCLC than in patients with limited disease (27 ). CgA plasma concentrations have also been found to correlate with

tumor weight in nude mice with xenografted human ileal carcinoid tumors (28 ). For some of the gastrointestinal neuroendocrine tumor patients with widespread, rapidly progressing disease the CgA PIA results seemed to enable better estimation of the tumor burden than the CgA(3403)assay results (results not shown). Further studies are needed to confirm this. CgA in plasma is frequently used as a marker for neuroendocrine tumors and prostatic carcinomas. Measurements are performed with antisera raised against purified full-length CgA or large fragments from tumor patients (3, 23, 29 –36 ), or with sequence-specific RIAs (15, 32, 37–39 ). Highly variable diagnostic sensitivities occur depending on tumor type and specificity of the assay employed. With a monoclonal antibody-based RIA used for patients with SCLC, increased concentrations were measured in 53% of patients with limited and 72% with extensive disease (40 ). In a study using a CgA ELISA for which the antiserum was raised against a C-terminal CgA fragment purified from urine of a carcinoid tumor patient, increased concentrations were measured in 37% of patients with SCLC (27 ). In the present study the corresponding results using the CgA(3403) assay were 100% (extensive disease) and 58% (limited disease), and using the CgA PIA 89% (extensive disease) and 67% (limited disease), respectively. In a study using an antiserum raised against fulllength CgA on plasma samples from 208 patients with neuroendocrine tumors, increased CgA concentrations were measured in 53% of the patients, most frequently observed in the pheochromocytoma (89%) and carcinoid tumor patients (80%) (29 ). In a similar study, plasma CgA was quantified with an RIA using an antiserum raised against C-terminal CgA fragments isolated from urine from a carcinoid tumor patient and a pancreastatin antiserum, respectively, and increased concentrations were measured in 99% with the CgA

444

Børglum et al.: Processing-Independent Assay for Chromogranin A

Table 4A. Diagnostic sensitivity obtained with the index test (CgA PIA) and 3 sequence-specific CgA assays for different tumor types.

Index test (CgA PIA)

CgA(3403) ⫹ trypsina

Sequence-specific CgA assays

CgA(13)

CgA(4301)

CgA(3403)

All tumors (n ⴝ 54)

Carcinoid (n ⴝ 22)

Other GI NE tumors (n ⴝ 5)

Pheochromocytoma (n ⴝ 6)

SCLC (n ⴝ 21)

0.82 (0.69–0.91) 44/54b 0.46 (0.33–0.60) 25/54 0.54 (0.40–0.67) 29/54 0.91 (0.80–0.97) 49/54

0.86 (0.65–0.97) 19/22 0.50 (0.28–0.72) 11/22 0.59 (0.36–0.79) 13/22 1.00 (0.85–1.00) 22/22

0.80 (0.28–1.00) 4/5 0.60 (0.15–0.95) 3/5 0.40 (0.05–0.85) 2/5 1.00 (0.48–1.00) 5/5

0.83 (0.36–1.00) 5/6 0.67 (0.22–0.96) 4/6 0.50 (0.12–0.88) 3/6 1.00 (0.54–1.00) 6/6

0.76 (0.53–0.92) 16/21 0.33 (0.15–0.60) 7/21 0.52 (0.30–0.74) 11/21 0.76 (0.53–0.92) 16/21

Table 4B. Diagnostic sensitivity obtained with the index test (CgA PIA) and 3 sequence-specific CgA assays, related to dissemination of disease. GI NE tumorsc

Index test (CgA PIA) Sequence-specific CgA assays

CgA(3403) ⫹ trypsina CgA(13) CgA(4301) CgA(3403)

SCLCd

Local (n ⴝ 3)

Extensive (n ⴝ 16)

Very extensive (n ⴝ 8)

Limited (n ⴝ 12)

Extensive (n ⴝ 9)

0.33 (0.01–0.91) 1/3b 0.00 (0.00–0.71) 0/3 0.33 (0.01–0.91) 1/3 1.00 (0.29–1.00) 3/3

0.88 (0.62–0.98) 14/16 0.38 (0.15–0.65) 6/16 0.56 (0.30–0.80) 9/16 1.00 (0.79–1.00) 16/16

1.00 (0.63–1.00) 8/8 1.00 (0.63–1.00) 8/8 0.63 (0.24–0.91) 5/8 1.00 (0.63–1.00) 8/8

0.67 (0.35–0.90) 8/12 0.17 (0.02–0.48) 2/12 0.42 (0.15–0.72) 5/12 0.58 (0.28–0.85) 7/12

0.89 (0.52–1.00) 8/9 0.55 (0.21–0.86) 5/9 0.67 (0.30–0.93) 6/9 1.00 (0.66–1.00) 9/9

a

Plasma samples were incubated with trypsin prior to CgA(3403) assay, as described in the text. Results are given as diagnostic sensitivity (95% CI) as well as the number of results above upper reference interval limit vs the total number of results. c GI NE (gastrointestinal neuroendocrine) tumors: The diagnostic sensitivities obtained with the CgA(13) assay and CgA PIA were significantly different for the local, extensive, and very extensive tumors (P ⫽ 0.002) and (P ⫽ 0.04). d SCLC: The diagnostic sensitivities obtained with the CgA(3403) assay were significantly different for the local and extensive SCLC (P ⫽ 0.045). b

assay and in 46% with the pancreastatin assay (32 ). In a study using an antiserum raised against CgA isolated from a pheochromocytoma, plasma CgA concentrations were increased in 81% of patients with neuroendocrine tumors (1 ). Furthermore, with use of an immunoradiometric assay increased concentrations were found in 63% of patients with gastroenteropancreatic endocrine tumors (26 ). Comparison of the diagnostic sensitivities obtained with the CgA assays referred to above and the results on the CgA(3403) assay and the CgA PIA reported in this study shows that both are particularly potent and sensitive tools in the diagnosis of neuroendocrine tumors and SCLC. Whether one of the 2 assays is superior to the other requires analysis on a larger patient population. Both the CgA(3403) assay and the CgA PIA results increased with the degree of metastatic disease and may thus reflect tumor burden. The diagnostic sensitivities generally increased with the degree of metastatic disease. However, for the CgA(3403) assay the sensitivity was very high in the patients with gastrointestinal tumors, irrespective of

dissemination. Further study is required to determine whether a low degree of CgA processing is a predictor of malignancy.

This study was supported by grants from the Danish Medical Research Council, the Danish Biotechnology Program for Cellular Communication, and the Danish Cancer Union. We thank Drs. Jens Gustafsen and Ulrich Knigge, Department of Surgical Gastroenterology, and Dr. Jørgen Kvist Kristensen, Department of Urology, Rigshospitalet, University of Copenhagen; Dr. Niels Juel Christensen, Department of Endocrinology, KAS Herlev University of Copenhagen for providing the tissue and plasma samples; and Jane Lancaster and Joan Christiansen for technical assistance. Dr. Ruth Frikke-Schmidt, Department of Clinical Biochemisty, Rigshospitalet and statisticians at the Department of Biostatistics, University of Copenhagen are thanked for statistical assistance.

Clinical Chemistry 53, No. 3, 2007

References 1. O’Connor DT, Deftos LJ. Secretion of chromogranin A by peptideproducing endocrine neoplasms. N Eng J Med 1986;314:1145– 51. 2. Konecki DS, Benedum UM, Gerdes HH, Huttner WB. The primary structure of human chromogranin A and pancreastatin. J Biol Chem 1987;262:17026 –30. 3. Sobol RE, Memoli V, Deftos LJ. Hormone-negative, chromogranin A-positive endocrine tumors. N Engl J Med 1989;320:444 –7. 4. Rosa P, Gerdes HH. The granin protein family. Markers for neuroendocrine cells and tools for the diagnosis for neuroendocrine tumors. J Endocrinol Invest 1994;17:207–25. 5. Rehfeld JF. Processing of precursors of gastroenteropancreatic hormones: diagnostic significance. J Mol Med 1998;76:338 – 45. 6. Bardram L. Gastrin in non-neoplastic pancreatic tissue from patients with and without gastrinomas. Scand J Gastroenterol 1990;25:935– 43. 7. Bardram L, Rehfeld JF. Processing-independent radioimmunoanalysis: a general analytical principle applied to progastrin and its products. Anal Biochem 1988;175:537– 43. 8. Bardram L. Progastrin in serum from Zollinger-Ellison patients. An indicator of malignancy? Gastroenterology 1990;98:1420 – 6. 9. Paloheimo LI, Rehfeld JF. Quantitation of procholecystokinin and its products in plasma by processing-independent analysis. Clin Chim Acta 1995;238:21–33. 10. Jørgensen NR, Rehfeld JF, Bardram L, Hilsted L. Processingindependent analysis in the diagnosis of gastrinomas. Scand J Gastroenterol 1998;33:379 – 85. 11. Goetze JP, Kastrup J, Pedersen F, Rehfeld JF. Quantification of pro-B-type natriuretic peptide and its products in human plasma by use of an analysis independent of precursor processing. Clin Chem 2002;48:1035– 42. 12. Goetze JP. Biochemistry of pro-B-type natriuretic peptide-derived peptides: the endocrine heart revisited. Clin Chem 2004;50: 1503–10. 13. Jais P, Mignon M, Rehfeld JF. Processing-independent assay of serum gastrin for diagnosis of liver metastases in the ZollingerEllison syndrome. Int J Cancer 1997;71:308 –9. 14. Reubi JC, Koefoed P, Hansen TO, Stauffer E, Rauch D, Nielsen FC, et al. Procholecystokinin as marker of human Ewing sarcomas. Clin Cancer Res 2004;10:5523–30. 15. Jensen TB, Hilsted L, Rehfeld JF. A library of sequence-specific radioimmunoassays for human chromogranin A. Clin Chem 1999; 45:549 – 60. 16. Børglum Jensen TD, Holst JJ, Fahrenkrug J. Characterization of immunoreactive pancreastatin in porcine tissues. Scand J Clin Lab Invest 1991;51:681–91. 17. Solcia E, Kloppel G, Sobin LH. Histological Typing of Endocrine Tumors, 2nd ed. New York: Springer, 2000. ¨ berg K, Stridsberg M. Tumor markers in neuroendo18. Eriksson B, O crine tumors. Digestion 2000;62(Suppl 1):33– 8. ¨ berg K, Astrup L, Eriksson B, Falkmer SE, Falkmer UG, Justafsen 19. O J, et al. Guidelines for the management of gastroenteropancreatic neuroendocrine tumours (including bronchopulmonary and thymic neoplasms). Acta Oncol 2004;43:617–25. 20. Travis WD, Colby TV, Corrin B, Shimosato Y, Brambilla E. Histological Typing of Lung and Pleural Tumours, 3rd ed. New York: Springer, 1999. 21. Tietz Textbook of Clinical Chemistry and Molecular Diagnostics, 4th ed. CA Burtis, ER Ashwood, DE Bruns, eds. St. Louis: Elsevier Saunders, 2006. 22. R Development Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2006 ISBN 3-900051-07-0. http:www.R-project. org (accessed January 2007).

445

23. Takiyyuddin MA, Cervenka JH, Pandian MR, Stuenkel CA, Neumann HP, O’Connor DT. Neuroendocrine sources of CgA in normal man: clues from selective stimulation of endocrine glands. J Clin Endocrinol Metab 1990;71:360 –9. 24. Jørgensen PE, Vinter-Jensen L, Nexø E. An immunoassay designed to quantitate different molecular forms of rat urinary epidermal growth factor with equimolar potency: application on fresh rat urine. Scand J Clin Lab Invest 1996;56:25–36. 25. Janson ET, Holmberg L, Stridsberg M, Eriksson B, Theodorsson, Wilander E, et al. Carcinoid tumors: analysis of prognostic factors and survival in 301 patients from a referral center. Annals of Oncology 1997;8:685–90. 26. Nehar D, Lombard-Bohas C, Olivieri S, Claustrat B, Chayvialle JA, Penes MC, et al. Interest of chromogranin A for diagnosis and follow-up of endocrine tumours. Clin Endocrinol (Oxf) 2004;60: 644 –52. 27. .Drivsholm L, Paloheimo LI, Osterlind K. Chromogranin A, a significant prognostic factor in small cell lung cancer. Br J Cancer 1999;81:667–71. 28. Ko¨lby L, Bernhardt P, Swa¨rd C, Johanson V, Ahlman H, ForssellAronsson E, et al. Chromogranin A as a determinant of midgut carcinoid tumour volume. Regul Pept 2004;120:269 –73. 29. Nobels FRE, Kwekkeboom DJ, Coopmans W, Schoenmakers CHH, Lindemans J, De Herder WW, et al. Chromogranin A as serum marker for neuroendocrine neoplasia: comparison with neuronspecific enolase and the ␣-subunit of glycoprotein hormones. J Clin Endocrinol Metab 1997;82:2622– 8. 30. Kimura N, Miura W, Noshiro T, Mizunashi K, Hanew K, Shimizu K, et al. Plasma chromogranin A in pheochromocytoma, primary hyperparathyroidism and pituitary adenoma in comparison with catecholamine, parathyroid hormone and pituitary hormones. Endocr J 1997;44:319 –27. 31. Stridsberg M, Oberg K, Li Q, Engstrøm U, Lundqvist G. Measurements of chromogranin A, chromogranin B (secretogranin I), chromogranin C (secretogranin II) and pancreastin in plasma and urine from patients with carcinoid tumours and endocrine pancreatic tumours. J Endocrinology 1995;144:49 –59. 32. Stridsberg M, Hellman U, Wilander E, Lundquist G, Hellsing K, Oberg K. Fragments of chromogranin A are present in the urine of patients with carcinoid tumors: development of a specific radioimmunoassay for chromogranin A and its fragments. J Endocrinol 1993;139:329 –37. 33. Schurmann G, Raeth U, Wiedenmann B, Buhr H, Herfarth C. Serum chromogranin A in the diagnosis and follow-up of neuroendocrine tumors of the gastroenteropancreatic tract. World J Surg 1992;16:697–702. 34. Syversen U, Jacobsen MB, O’Connor DT, Rønning K, Waldum HL. Immunoassays for measurement of chromogranin A and pancreastatin-like immunoreactivity in humans: correspondence in patients with neuroendocrine neoplasia. Neuropeptides 1994;26: 201– 6. 35. Deftos LJ, Nakada S, Burton DW, di Sant’Agnese PA, Cockett ATK, Abrahamsson PA. Immunoassay and immunohistology studies of chromogranin A as a neuroendocrine marker in patients with carcinoma of the prostate. Urology 1996;48:58 – 62. 36. Deftos LJ, Abrahamsson PA. Granins and prostate cancer. Urology 1998;51(Suppl 5A):141–5. 37. McGrath-Linden SJ, Johnston CF, O’Connor DT, Buchanan KD. Pancreastatin-like immunoreactivity in human carcinoid disease. Regul Pept 1991;33:55–70.

446

Børglum et al.: Processing-Independent Assay for Chromogranin A

¨ berg K, Janson ET. A panel of 11 38. Stridsberg M, Eriksson B, O region-specific radioimmunoassays for measurements of chromogranin A. Regul Pept 2004;117:219 –27. ¨ berg K, Vezzadini P, Foschini MP, 39. Tartaglia A, Portela-Gomes GM, O Stridsberg M. Chromogranin A in gastric neuroendocrine tumours:

an immunohistochemical and biochemical study with regionspecific antibodies. Virchows Arch 2000;12:1– 8. 40. Sobol RE, O’Connor DT, Addison J, Suchocki K, Reyston I, Deftos L. Elevated serum chromogranin A concentrations in small-cell lung carcinoma. Ann Intern Med 1986;105:698 –700.