Med Oncol (2014) 31:823 DOI 10.1007/s12032-013-0823-1
ORIGINAL PAPER
The prognostic significance of the circulating neuroendocrine markers chromogranin A, pro-gastrin-releasing peptide, and neuron-specific enolase in patients with small-cell lung cancer Marina Petrovic´ • Zoran Bukumiric´ • Vladimir Zdravkovic´ • Slobodanka Mitrovic´ Henry Dushan Atkinson • Vladimir Jurisˇic´
•
Received: 3 September 2013 / Accepted: 17 December 2013 / Published online: 30 December 2013 Ó Springer Science+Business Media New York 2013
Abstract Lung cancer is the most common cancer, and small-cell lung cancer (SCLC) accounts for around 20 % of lung cancers. SCLC has a neuroendocrine cellular origin, and the tumor cells usually express neuroendocrine markers. There have been major recent advances in the management of SCLC, and multimodal approaches are now the norm. An improved knowledge of the prognostic variables would assist in defining which patients were better candidates to receive these newer intensive therapies. This single-center retrospective study of 97 previously untreated and histologically proven SCLC patients analysed the circulating neuroendocrine markers chromogranin A (CGA), pro-gastrin-releasing peptide (ProGRP), and neuron-specific enolase (NSE) in addition to the other more classical variables. Fifty patients had limitedstage disease and 47 had extensive disease. Sixty patients had an ECOG performance status (PS) of 0–1 and 37 had PS 2–4. Median survival for the whole study population was 13 months. Univariate analysis and univariate Cox regression modeling found a statistically significant association between survival and PS, disease stage, and CGA, ProGRP, and NSE levels. Age and sex were not prognostic. A shorter survival time was found in patients with a PS equal to or [2, extensive stage disease, a serum CGA level
M. Petrovic´ V. Zdravkovic´ S. Mitrovic´ V. Jurisˇic´ (&) Faculty of Medical Sciences, University of Kragujevac, Kragujevac, Serbia e-mail: [email protected] Z. Bukumiric´ Faculty of Medicine, University of Belgrade, Belgrade, Serbia H. D. Atkinson North Middlesex University Hospital, Sterling Way, London, UK
[56 ng/ml, a serum ProGRP level[58 pg/ml, and a serum NSE level [19 ng/ml. This study has found that there is a potential role for ProGRP, NSE, and CGA in both staging and prognosing survival in SCLC patients. Keywords Lung cancer SCLC ELISA Prognosis Therapy Neuroendocrine markers
Introduction Lung cancer is the most common worldwide malignancy. It is classified into small cell lung cancer (SCLC) (20 %) and non-small cell lung cancer (NSCLC) (80 %). SCLC is characterized by its rapid doubling time and propensity for early metastases. In essence, there are two clinical stages, limited disease (LD), where the tumor is confined to one hemithorax, and extensive disease (ED), when metastases occur in the contralateral chest and at distant sites. Metastases initially occur in the lymph nodes and thereafter in other organs such as other areas of the lung, the liver, adrenal glands, brain, bone, and bone marrow [1, 2]. Twenty to 25 % of patients have LD, and though treatment is potentially curative, 5-year survival rates are poor (15–25 %, compared with \5 % in ED patients); and median survival (MS) times are 14–20 months and 7–10 months in LD and ED patients, respectively [3]. There have been major advances in the management of SCLC over the past two decades, and multimodal approaches are now the norm [4]. One can achieve an 80–90 % overall response rate with a 30–40 % complete response rate by using several cytotoxic agents. The treatment of choice for ED SCLC is currently cisplatin or carboplatin and etoposide. Some protocols also include newer drugs such as topoisomerase I inhibitors and taxans [1, 5, 6].
123
823 Page 2 of 7
Radiotherapy is frequently given in combination with chemotherapy; additional prophylactic cranial irradiation has also been suggested to help prevent cerebral metastases; though the optimal timing, dosage and fractionation of radiotherapy have yet to be fully defined [1, 5, 6]. Despite the high chemosensitivity of SCLC, the 2- and 5-year survival remains low due to frequent chemoresistant relapses. As a result different strategies are now being evaluated in order to circumvent this secondary chemoresistance and to improve survival [7]. The three main approaches involve using new drug combinations with newer drugs [8–10], the use of adjunct biological response modifiers, and the use of chemotherapy intensification [11, 12]; though concerns have been raised regarding these strategies and patient eligibility criteria [13]. Thus a better knowledge of the prognostic variables would allow us to more clearly define which patients were better candidates to receive these newer intensive therapies. Until recently, the main prognostic indicators for SCLC were the patient’s performance status and the disease stage [14, 15]. SCLC has a neuroendocrine cellular origin and its tumor cells have a similar antigenic profile; they often express neuroendocrine markers. Though neuroendocrine differentiation is considered to influence the clinical behavior of SCLC, these serum neuroendocrine markers have not yet been routinely used as prognostic indicators in clinical practice. Potentially, useful neuroendocrine disease markers include chromogranin A (CGA), pro-gastrinreleasing peptide (ProGRP), neuron-specific enolase (NSE; an c–c isoform of the ubiquitous enolase enzyme), cytokeratin 19 marker CYFRA 21–1, synaptophysin, creatine kinase BB, and neural cell adhesion molecule [4]. CGA is a 49 kDa acidic-soluble protein ubiquitous in neuroendocrine tissues, and elevated levels have been reported in patients suffering from SCLC. Gastrin-releasing peptide (GRP) is a gut hormone and is present in nerve fibers, the brain and in neuroendocrine cells in the fetal lung [16]. It is commonly elevated in SCLC patients, however has not been used as a diagnostic marker as it is quite unstable in the serum [17]. Various studies have shown that its precursor ProGRP [17–29] is superior to other markers in its ability to differentiate between SCLC and NSCLC, and serum ProGRP levels can be measured by ELISA. ProGRP and GRPR mRNA expression can also be measured in SCLC tumor tissues through reverse transcription-nested polymerase chain reaction (RT-PCR) amplification [18]. These transcriptions do not occur in normal non-neoplastic cells, and therefore RT-PCR can assist us in the earlier detection of rare disseminated or exfoliated cancer cells in the peripheral venous blood and the sputum of lung cancer patients; thus helping us to more accurately stage the disease. RT-PCR assays have been used to identify various neuroendocrine marker transcripts
123
Med Oncol (2014) 31:823
in this way including neural cell adhesion molecule (NCAM), protein gene product (PGP 9.5), gastrin, gastrin receptors, synaptophysin, prepro-gastrin-releasing peptide (preproGRP), and GRP-receptor [19]. This retrospective study analyses the neuroendocrine markers CGA, ProGRP, and NSE in addition to the other more classical variables and aims to determine whether they might hold a prognostic role in SCLC.
Patients and methods A total of 97 histologically proven and previously untreated SCLC patients were included in the study. All the patients underwent clinical examination, chest X-ray, CT scans of the chest and brain, CT and ultrasonography of the upper abdomen, fibreoptic bronchoscopy, bone scanning, and unilateral bone marrow biopsy. Staging was carried out according to the veterans administration lung cancer group (VALG) staging system [20]: Limited disease was defined as disease confined to one hemithorax including the mediastinal lymph nodes and/or the supraclavicular lymph nodes; extensive disease was defined in those not having limited disease and in patients with a malignant pleural effusion. The study was approved by the Local Ethics Committee. Treatments All patients were treated with cisplatin–etoposide chemotherapy for a maximum of six cycles. Forty-three LD patients also received thoracic radiotherapy with an alternating chemo-radiotherapy schedule. Fourteen patients received supportive care (including palliative radiotherapy for bony and brain metastases); eight of these had a performance status [2, while six had a performance status equal to 2, an age over 70 years and major co-morbidities. Biochemical measurements After obtaining informed consent, serum samples were taken from all patients before the start of chemotherapy and stored at -80 °C. Serum CGA and serum NSE were measured using commercially available solid-phase ‘‘two site sandwich’’ immunoradiometric assays (CGA-RIACT and ELSANSE Cis Bio International, Gif-sur-Yvette, France). The CGA-RIACT assay involved two monoclonal antibodies raised against the amino acid sequences 145–197 and 198–245 of human CGA. The assay detected both intact molecules and different molecular fragments. The concentration of CGA was expressed in ng/ml. The ELSA-NSE assay explored two monoclonal antibodies prepared against sterically remote antigenic sites on the NSE molecule. The first was specific for the NSE coated on the ELSA solid
Med Oncol (2014) 31:823
phase, and the second was radiolabelled with iodine-125 and was used as a tracer. The concentration of NSE was expressed in ng/ml. Serum ProGRP was measured with the ELISA kit manufactured by ALSI, Japan, and distributed in Europe by IBL-Hamburg, Germany. The assay followed the principle of a two-step sandwich-type assay. The ProGRP concentration was expressed in pg/ml. All tests, including hemoglobin (Hb), serum creatinine, lactate dehydrogenase, alkaline phosphatase, leukocytes, albumin, were performed before the initiation of chemotherapy. Statistical analyses Survival was measured from the start of chemotherapy to the latest follow-up evaluation or death. Univariate survival analyses of the impact of various prognostic factors were performed using the Kaplan–Meier method and the logrank test. Some of the variables have been extensively described in the literature, and thresholds were defined by previous publications. Performance status was analysed according to the two classical groupings: PS 0–1 and PS C2 [21]. The effects of the disease were tested according to the two-group staging system (limited and extensive disease). The age at diagnosis and any survival relationship was evaluated according to the grouping: equal to or younger than 62 years, and older than 62 years; 62 years was the median age in our series. We used thresholds defined by our previous publication when analysing biological variables and tumor markers; the serum marker thresholds were determined according to the maximum Youden indices. For ProGRP, the threshold value of 58 pg/ ml resulted in the maximum indices and produced a 0.84 sensitivity and a 0.92 specificity; similar methodology determined a 19 ng/ml threshold for NSE with a 0.80 sensitivity and a 0.95 specificity; a threshold of 56 ng/ml for CgA with a 0.74 sensitivity and a 0.82 specificity. Variables with a p value \ 0.05 were entered into multivariate Cox regression models. Associations between categorical variables were evaluated with Fisher’s exact t test or v2 test. Correlations between numerical variables were analyzed by a linear nonparametric (Spearman) correlation test. Differences in the numerical variables between categorical factors were tested using the Mann–Whitney test. Statistical calculations were performed using SPSS for Windows V10. A p value\0.05 was considered significant.
Page 3 of 7 823
patients had limited-stage disease and 47 had extensive disease. Sixty patients had an ECOG PS 0–1 and 37 had PS 2–4. Median follow-up was 13 (range 1–31) months. Tumor marker distribution by age, gender, stage, and PS Table 1 shows the distribution of serum levels of CGA, ProGRP, and NSE in SCLC patients. No association was found between the varying serum levels of NSE, CGA, ProGRP with age or gender. However, these levels varied significantly with PS and disease extent. The median serum levels of CGA (69 vs 32.5 ng/ml; p \ 0.001), ProGRP (158 vs 51 pg/ml; p \ 0.001), and NSE (28.6 vs 15.5.ng/ ml; p \ 0.001) were significantly greater in patients with a PS of 2–4 compared with patients with a PS of 0–1. Those with extensive disease also had significantly greater marker levels than those with limited disease (CGA 68 vs 27 ng/ ml; p \ 0.001, ProGRP 123 vs 48.5 pg/ml; p \ 0.001, NSE 32.4 vs 14 ng/ml; p \ 0.001, respectively). Univariate and multivariate survival analyses Median survival for the whole study population was 13 months. Univariate analysis found a statistically significant association of survival with PS (p = 0.001), disease stage (p \ 0.001), CGA (p \ 0.001), ProGRP (p \ 0.001), and NSE (p \ 0.001). Age and sex were not prognostic (Table 2). Univariate Cox regression modeling found that PS (B = -0.7; p = 0.002), disease stage (B = -1.12; p \ 0.001), CGA (B = 0.004; p = 0.001), ProGRP (B = 0.001; p = 0.016), and NSE (B = 0.005; p = 0.008) were statistically significant predictors for survival (Table 3). Univariate analysis found that the following variables conferred a shorter survival time when compared to their opposing ‘‘arms’’: a performance status equal to or worse than 2 (MS of 12 vs 14 months; p = 0.001 Fig. 1), extensive stage disease (MS 10 vs 15 months; p \ 0.001 Fig. 2), a serum CGA level [56 ng/ml (MS 10 vs 14 months; p \ 0.001 Fig. 3), a serum ProGRP level[58 pg/ml (MS 10 vs 15 months; p \ 0.001 Fig. 4), and a serum NSE level [19 ng/ml (MS 10 vs 15 months; p \ 0001 Fig. 5). All the significant variables were then analyzed (Table 4), using the multiple Cox regression model, which found that performance status and disease extent were significant predictors (B = 0.824; p = 0.001 and B = –1.19; p \ 0.001, respectively).
Results Discussion Patient characteristics The study included 97 patients (62 males/35 females) with a median age of 62 years (range 37–78 years). Fifty
SCLC is a malignancy which is associated with neuroendocrine differentiation; thus, neuroendocrine markers such as NSE, chromogranin A, synaptophysin, creatinine kinase
123
823 Page 4 of 7
Med Oncol (2014) 31:823
Table 1 Distribution of serum levels of CGA, ProGRP, and NSE in SCLC according to age, gender, PS, and disease stage Factor
n
Age
CGA (ng/ml) mean ± SD; range; median
ProGRP (pg/ml) mean ± SD; range; median
NSE (ng/ml) mean ± SD; range; median
p = 0.716
p = 0.454
p = 0997
B62 years
43
63.9 ± 68.24; 17–421; 48
111.53 ± 158.82; 12–967; 55
35.3 ± 46.45; 5.9–232.8; 19.7
[62 years
54
73.98 ± 96.2; 14–479; 42.5
149.4 ± 235.78; 17–1,359; 67.5
36.42 ± 58.92; 3.8–324.1; 21.25
p = 0.353
p = 0.462
p = 0.142
Male
62
60.48 ± 67.9; 15–479; 39.5
110.19 ± 139.86; 12–931; 55.5
29.66 ± 37.31; 3.8–232.8; 18
Female
35
85.54 ± 107.54; 14–451; 49
172.3 ± 284.95; 17–1,359; 65
47.02 ± 73.36; 5.6–324.1; 22.7
PS 0–1
69
p < 0.001 50.16 ± 64.65; 14–421; 32.5
p < 0.001 72.92 ± 98.15; 12–754; 51
p < 0.001 20.1 ± 14.6; 3.8–88.5; 15.5
2–4
28
Sex
Stage
100.92 ± 103.15; 23–479; 69
229.43 ± 284.36; 32–1,359; 158
61.58 ± 78.66; 5.6–324.1; 28.6
p < 0.001
p < 0.001
p < 0.001
Limited
50
35.98 ± 21.23; 14–89; 27
58.96 ± 41.14; 12–179; 48.5
14.99 ± 6.57; 3.8–34.4; 14
Extensive
47
105.21 ± 109.38; 26–479; 68
210.97 ± 271.76; 21–1,359; 123
58.19 ± 70.3; 5.6–324.1; 32.4
Bold values indicate statistically significant values (p \ 0.5) Table 2 Kaplan–Meier univariate survival analysis of the prognostic factors Factor
Patients (n)
Median survival (months)
95 % confidence interval
p value logrank
Age B62 years
43
13
10.67–15.3
[62 years
54
13
11.53–14.47
0.377
Sex Male
62
13
11.68–14.32
Female
35
12
11.08–12.92
0–1
60
14
12.41–15.59
2–4
37
12
10.01–13.99 13.9–16.1
0.446
Table 3 Univariate Cox regression modeling Factor
Hazard ratio
95 % confidence interval
p value log-rank
Age
1.02
0.99–1.05
0.106
Sex
1.17
0.75–1.85
0.481
PS
0.49
0.31–0.77
0.002
Stage
0.33
0.21–0.52
ORIGINAL PAPER
The prognostic significance of the circulating neuroendocrine markers chromogranin A, pro-gastrin-releasing peptide, and neuron-specific enolase in patients with small-cell lung cancer Marina Petrovic´ • Zoran Bukumiric´ • Vladimir Zdravkovic´ • Slobodanka Mitrovic´ Henry Dushan Atkinson • Vladimir Jurisˇic´
•
Received: 3 September 2013 / Accepted: 17 December 2013 / Published online: 30 December 2013 Ó Springer Science+Business Media New York 2013
Abstract Lung cancer is the most common cancer, and small-cell lung cancer (SCLC) accounts for around 20 % of lung cancers. SCLC has a neuroendocrine cellular origin, and the tumor cells usually express neuroendocrine markers. There have been major recent advances in the management of SCLC, and multimodal approaches are now the norm. An improved knowledge of the prognostic variables would assist in defining which patients were better candidates to receive these newer intensive therapies. This single-center retrospective study of 97 previously untreated and histologically proven SCLC patients analysed the circulating neuroendocrine markers chromogranin A (CGA), pro-gastrin-releasing peptide (ProGRP), and neuron-specific enolase (NSE) in addition to the other more classical variables. Fifty patients had limitedstage disease and 47 had extensive disease. Sixty patients had an ECOG performance status (PS) of 0–1 and 37 had PS 2–4. Median survival for the whole study population was 13 months. Univariate analysis and univariate Cox regression modeling found a statistically significant association between survival and PS, disease stage, and CGA, ProGRP, and NSE levels. Age and sex were not prognostic. A shorter survival time was found in patients with a PS equal to or [2, extensive stage disease, a serum CGA level
M. Petrovic´ V. Zdravkovic´ S. Mitrovic´ V. Jurisˇic´ (&) Faculty of Medical Sciences, University of Kragujevac, Kragujevac, Serbia e-mail: [email protected] Z. Bukumiric´ Faculty of Medicine, University of Belgrade, Belgrade, Serbia H. D. Atkinson North Middlesex University Hospital, Sterling Way, London, UK
[56 ng/ml, a serum ProGRP level[58 pg/ml, and a serum NSE level [19 ng/ml. This study has found that there is a potential role for ProGRP, NSE, and CGA in both staging and prognosing survival in SCLC patients. Keywords Lung cancer SCLC ELISA Prognosis Therapy Neuroendocrine markers
Introduction Lung cancer is the most common worldwide malignancy. It is classified into small cell lung cancer (SCLC) (20 %) and non-small cell lung cancer (NSCLC) (80 %). SCLC is characterized by its rapid doubling time and propensity for early metastases. In essence, there are two clinical stages, limited disease (LD), where the tumor is confined to one hemithorax, and extensive disease (ED), when metastases occur in the contralateral chest and at distant sites. Metastases initially occur in the lymph nodes and thereafter in other organs such as other areas of the lung, the liver, adrenal glands, brain, bone, and bone marrow [1, 2]. Twenty to 25 % of patients have LD, and though treatment is potentially curative, 5-year survival rates are poor (15–25 %, compared with \5 % in ED patients); and median survival (MS) times are 14–20 months and 7–10 months in LD and ED patients, respectively [3]. There have been major advances in the management of SCLC over the past two decades, and multimodal approaches are now the norm [4]. One can achieve an 80–90 % overall response rate with a 30–40 % complete response rate by using several cytotoxic agents. The treatment of choice for ED SCLC is currently cisplatin or carboplatin and etoposide. Some protocols also include newer drugs such as topoisomerase I inhibitors and taxans [1, 5, 6].
123
823 Page 2 of 7
Radiotherapy is frequently given in combination with chemotherapy; additional prophylactic cranial irradiation has also been suggested to help prevent cerebral metastases; though the optimal timing, dosage and fractionation of radiotherapy have yet to be fully defined [1, 5, 6]. Despite the high chemosensitivity of SCLC, the 2- and 5-year survival remains low due to frequent chemoresistant relapses. As a result different strategies are now being evaluated in order to circumvent this secondary chemoresistance and to improve survival [7]. The three main approaches involve using new drug combinations with newer drugs [8–10], the use of adjunct biological response modifiers, and the use of chemotherapy intensification [11, 12]; though concerns have been raised regarding these strategies and patient eligibility criteria [13]. Thus a better knowledge of the prognostic variables would allow us to more clearly define which patients were better candidates to receive these newer intensive therapies. Until recently, the main prognostic indicators for SCLC were the patient’s performance status and the disease stage [14, 15]. SCLC has a neuroendocrine cellular origin and its tumor cells have a similar antigenic profile; they often express neuroendocrine markers. Though neuroendocrine differentiation is considered to influence the clinical behavior of SCLC, these serum neuroendocrine markers have not yet been routinely used as prognostic indicators in clinical practice. Potentially, useful neuroendocrine disease markers include chromogranin A (CGA), pro-gastrinreleasing peptide (ProGRP), neuron-specific enolase (NSE; an c–c isoform of the ubiquitous enolase enzyme), cytokeratin 19 marker CYFRA 21–1, synaptophysin, creatine kinase BB, and neural cell adhesion molecule [4]. CGA is a 49 kDa acidic-soluble protein ubiquitous in neuroendocrine tissues, and elevated levels have been reported in patients suffering from SCLC. Gastrin-releasing peptide (GRP) is a gut hormone and is present in nerve fibers, the brain and in neuroendocrine cells in the fetal lung [16]. It is commonly elevated in SCLC patients, however has not been used as a diagnostic marker as it is quite unstable in the serum [17]. Various studies have shown that its precursor ProGRP [17–29] is superior to other markers in its ability to differentiate between SCLC and NSCLC, and serum ProGRP levels can be measured by ELISA. ProGRP and GRPR mRNA expression can also be measured in SCLC tumor tissues through reverse transcription-nested polymerase chain reaction (RT-PCR) amplification [18]. These transcriptions do not occur in normal non-neoplastic cells, and therefore RT-PCR can assist us in the earlier detection of rare disseminated or exfoliated cancer cells in the peripheral venous blood and the sputum of lung cancer patients; thus helping us to more accurately stage the disease. RT-PCR assays have been used to identify various neuroendocrine marker transcripts
123
Med Oncol (2014) 31:823
in this way including neural cell adhesion molecule (NCAM), protein gene product (PGP 9.5), gastrin, gastrin receptors, synaptophysin, prepro-gastrin-releasing peptide (preproGRP), and GRP-receptor [19]. This retrospective study analyses the neuroendocrine markers CGA, ProGRP, and NSE in addition to the other more classical variables and aims to determine whether they might hold a prognostic role in SCLC.
Patients and methods A total of 97 histologically proven and previously untreated SCLC patients were included in the study. All the patients underwent clinical examination, chest X-ray, CT scans of the chest and brain, CT and ultrasonography of the upper abdomen, fibreoptic bronchoscopy, bone scanning, and unilateral bone marrow biopsy. Staging was carried out according to the veterans administration lung cancer group (VALG) staging system [20]: Limited disease was defined as disease confined to one hemithorax including the mediastinal lymph nodes and/or the supraclavicular lymph nodes; extensive disease was defined in those not having limited disease and in patients with a malignant pleural effusion. The study was approved by the Local Ethics Committee. Treatments All patients were treated with cisplatin–etoposide chemotherapy for a maximum of six cycles. Forty-three LD patients also received thoracic radiotherapy with an alternating chemo-radiotherapy schedule. Fourteen patients received supportive care (including palliative radiotherapy for bony and brain metastases); eight of these had a performance status [2, while six had a performance status equal to 2, an age over 70 years and major co-morbidities. Biochemical measurements After obtaining informed consent, serum samples were taken from all patients before the start of chemotherapy and stored at -80 °C. Serum CGA and serum NSE were measured using commercially available solid-phase ‘‘two site sandwich’’ immunoradiometric assays (CGA-RIACT and ELSANSE Cis Bio International, Gif-sur-Yvette, France). The CGA-RIACT assay involved two monoclonal antibodies raised against the amino acid sequences 145–197 and 198–245 of human CGA. The assay detected both intact molecules and different molecular fragments. The concentration of CGA was expressed in ng/ml. The ELSA-NSE assay explored two monoclonal antibodies prepared against sterically remote antigenic sites on the NSE molecule. The first was specific for the NSE coated on the ELSA solid
Med Oncol (2014) 31:823
phase, and the second was radiolabelled with iodine-125 and was used as a tracer. The concentration of NSE was expressed in ng/ml. Serum ProGRP was measured with the ELISA kit manufactured by ALSI, Japan, and distributed in Europe by IBL-Hamburg, Germany. The assay followed the principle of a two-step sandwich-type assay. The ProGRP concentration was expressed in pg/ml. All tests, including hemoglobin (Hb), serum creatinine, lactate dehydrogenase, alkaline phosphatase, leukocytes, albumin, were performed before the initiation of chemotherapy. Statistical analyses Survival was measured from the start of chemotherapy to the latest follow-up evaluation or death. Univariate survival analyses of the impact of various prognostic factors were performed using the Kaplan–Meier method and the logrank test. Some of the variables have been extensively described in the literature, and thresholds were defined by previous publications. Performance status was analysed according to the two classical groupings: PS 0–1 and PS C2 [21]. The effects of the disease were tested according to the two-group staging system (limited and extensive disease). The age at diagnosis and any survival relationship was evaluated according to the grouping: equal to or younger than 62 years, and older than 62 years; 62 years was the median age in our series. We used thresholds defined by our previous publication when analysing biological variables and tumor markers; the serum marker thresholds were determined according to the maximum Youden indices. For ProGRP, the threshold value of 58 pg/ ml resulted in the maximum indices and produced a 0.84 sensitivity and a 0.92 specificity; similar methodology determined a 19 ng/ml threshold for NSE with a 0.80 sensitivity and a 0.95 specificity; a threshold of 56 ng/ml for CgA with a 0.74 sensitivity and a 0.82 specificity. Variables with a p value \ 0.05 were entered into multivariate Cox regression models. Associations between categorical variables were evaluated with Fisher’s exact t test or v2 test. Correlations between numerical variables were analyzed by a linear nonparametric (Spearman) correlation test. Differences in the numerical variables between categorical factors were tested using the Mann–Whitney test. Statistical calculations were performed using SPSS for Windows V10. A p value\0.05 was considered significant.
Page 3 of 7 823
patients had limited-stage disease and 47 had extensive disease. Sixty patients had an ECOG PS 0–1 and 37 had PS 2–4. Median follow-up was 13 (range 1–31) months. Tumor marker distribution by age, gender, stage, and PS Table 1 shows the distribution of serum levels of CGA, ProGRP, and NSE in SCLC patients. No association was found between the varying serum levels of NSE, CGA, ProGRP with age or gender. However, these levels varied significantly with PS and disease extent. The median serum levels of CGA (69 vs 32.5 ng/ml; p \ 0.001), ProGRP (158 vs 51 pg/ml; p \ 0.001), and NSE (28.6 vs 15.5.ng/ ml; p \ 0.001) were significantly greater in patients with a PS of 2–4 compared with patients with a PS of 0–1. Those with extensive disease also had significantly greater marker levels than those with limited disease (CGA 68 vs 27 ng/ ml; p \ 0.001, ProGRP 123 vs 48.5 pg/ml; p \ 0.001, NSE 32.4 vs 14 ng/ml; p \ 0.001, respectively). Univariate and multivariate survival analyses Median survival for the whole study population was 13 months. Univariate analysis found a statistically significant association of survival with PS (p = 0.001), disease stage (p \ 0.001), CGA (p \ 0.001), ProGRP (p \ 0.001), and NSE (p \ 0.001). Age and sex were not prognostic (Table 2). Univariate Cox regression modeling found that PS (B = -0.7; p = 0.002), disease stage (B = -1.12; p \ 0.001), CGA (B = 0.004; p = 0.001), ProGRP (B = 0.001; p = 0.016), and NSE (B = 0.005; p = 0.008) were statistically significant predictors for survival (Table 3). Univariate analysis found that the following variables conferred a shorter survival time when compared to their opposing ‘‘arms’’: a performance status equal to or worse than 2 (MS of 12 vs 14 months; p = 0.001 Fig. 1), extensive stage disease (MS 10 vs 15 months; p \ 0.001 Fig. 2), a serum CGA level [56 ng/ml (MS 10 vs 14 months; p \ 0.001 Fig. 3), a serum ProGRP level[58 pg/ml (MS 10 vs 15 months; p \ 0.001 Fig. 4), and a serum NSE level [19 ng/ml (MS 10 vs 15 months; p \ 0001 Fig. 5). All the significant variables were then analyzed (Table 4), using the multiple Cox regression model, which found that performance status and disease extent were significant predictors (B = 0.824; p = 0.001 and B = –1.19; p \ 0.001, respectively).
Results Discussion Patient characteristics The study included 97 patients (62 males/35 females) with a median age of 62 years (range 37–78 years). Fifty
SCLC is a malignancy which is associated with neuroendocrine differentiation; thus, neuroendocrine markers such as NSE, chromogranin A, synaptophysin, creatinine kinase
123
823 Page 4 of 7
Med Oncol (2014) 31:823
Table 1 Distribution of serum levels of CGA, ProGRP, and NSE in SCLC according to age, gender, PS, and disease stage Factor
n
Age
CGA (ng/ml) mean ± SD; range; median
ProGRP (pg/ml) mean ± SD; range; median
NSE (ng/ml) mean ± SD; range; median
p = 0.716
p = 0.454
p = 0997
B62 years
43
63.9 ± 68.24; 17–421; 48
111.53 ± 158.82; 12–967; 55
35.3 ± 46.45; 5.9–232.8; 19.7
[62 years
54
73.98 ± 96.2; 14–479; 42.5
149.4 ± 235.78; 17–1,359; 67.5
36.42 ± 58.92; 3.8–324.1; 21.25
p = 0.353
p = 0.462
p = 0.142
Male
62
60.48 ± 67.9; 15–479; 39.5
110.19 ± 139.86; 12–931; 55.5
29.66 ± 37.31; 3.8–232.8; 18
Female
35
85.54 ± 107.54; 14–451; 49
172.3 ± 284.95; 17–1,359; 65
47.02 ± 73.36; 5.6–324.1; 22.7
PS 0–1
69
p < 0.001 50.16 ± 64.65; 14–421; 32.5
p < 0.001 72.92 ± 98.15; 12–754; 51
p < 0.001 20.1 ± 14.6; 3.8–88.5; 15.5
2–4
28
Sex
Stage
100.92 ± 103.15; 23–479; 69
229.43 ± 284.36; 32–1,359; 158
61.58 ± 78.66; 5.6–324.1; 28.6
p < 0.001
p < 0.001
p < 0.001
Limited
50
35.98 ± 21.23; 14–89; 27
58.96 ± 41.14; 12–179; 48.5
14.99 ± 6.57; 3.8–34.4; 14
Extensive
47
105.21 ± 109.38; 26–479; 68
210.97 ± 271.76; 21–1,359; 123
58.19 ± 70.3; 5.6–324.1; 32.4
Bold values indicate statistically significant values (p \ 0.5) Table 2 Kaplan–Meier univariate survival analysis of the prognostic factors Factor
Patients (n)
Median survival (months)
95 % confidence interval
p value logrank
Age B62 years
43
13
10.67–15.3
[62 years
54
13
11.53–14.47
0.377
Sex Male
62
13
11.68–14.32
Female
35
12
11.08–12.92
0–1
60
14
12.41–15.59
2–4
37
12
10.01–13.99 13.9–16.1
0.446
Table 3 Univariate Cox regression modeling Factor
Hazard ratio
95 % confidence interval
p value log-rank
Age
1.02
0.99–1.05
0.106
Sex
1.17
0.75–1.85
0.481
PS
0.49
0.31–0.77
0.002
Stage
0.33
0.21–0.52
