- Email: [email protected]
Pre-operative plasma erythropoietin concentration and survival following surgery for non-small cell lung cancer
Lung Cancer (2006) 51, 329—334
Pre-operative plasma erythropoietin concentration and survival following surgery for non-small cell lung cancer I. Paul a,∗, T.R.J. Lappin b, P. Maxwell c, A.N.J. Graham a a
Department of Cardiothoracic Surgery, Royal Victoria Hospital, Northern Ireland, Belfast, UK Department of Haematology, Belfast City Hospital, Northern Ireland, Belfast, UK c Department of Pathology, Queen’s University Belfast, Institute of Clinical Science, The Royal Group of Hospitals, Northern Ireland, Belfast, UK b
Received 10 June 2005 ; received in revised form 10 October 2005; accepted 31 October 2005 KEYWORDS Non-small cell lung carcinoma; Erythropoietin; Prognosis; Stage; Resectability; Haemoglobin
Summary Background: Suppression of the effect of the hormone erythropoietin (EPO) on the bone marrow, and an inadequate EPO response to anaemia have been shown to be factors in the genesis of cancer related anaemia. Low haemoglobin (Hb) concentration pre-operatively has been shown to have prognostic significance in patients with surgically resected NSCLC. This study investigates the relationship between pre-operative EPO and survival in patients having surgery for NSCLC. Methods: Pre-operative plasma EPO concentration and haemoglobin concentration were analysed in patients undergoing surgery for NSCLC between April 1998 and January 1999. Full follow-up was available for all patients. Results: Forty two patients were included. Median EPO concentration was 9.4 mIU/ml, range (3.7—56.4) with 17 patients (40.4%) having values above the normal range. Median haemoglobin concentration was 13.3 g/dl (range 8.5—16.8) with 15 patients (26%) anaemic pre-operatively. Pathological staging revealed 17 (40.4%) patients with stage I, 6 (14.3%) with stage II, 19 (45.3%) with stage III disease. Ten patients had irresectable disease. There was a significant difference in median EPO but not haemoglobin concentration, between the different pathological stages. Survival was significantly lower in patients with pre-operative EPO >10.5 mIU. Conclusions: Raised pre-operative EPO is associated with reduced survival in patients having surgery for NSCLC. Its measurement should be considered in the pre-operative
∗
Corresponding author at: Department of Cardiothoracic Surgery, Royal Victoria Hospital, Grosvenor Road, Northern Ireland, Belfast BT12 6BA, UK. Tel.: +44 28 9024 0503; fax: +44 28 9024 0503. E-mail address: [email protected] (I. Paul). 0169-5002/$ — see front matter © 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.lungcan.2005.10.020
330
I. Paul et al. assessment of patients undergoing surgery for NSCLC. Further research is required to further investigate the biological relationship between EPO and NSCLC. © 2005 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Anaemia is a common finding in patients with solid tumours, including non-small cell lung carcinoma (NSCLC). Factors involved in the pathogenesis of the anaemia of cancer are multiple. Haemodilution, bleeding from tumour erosion, hypersplenism, haemolysis, nutritional deficiencies, bone marrow infiltration, chemotherapy and radiotherapy have all been highlighted as contributing factors [1]. Many of these causes usually do not develop early in the course of the disease, when surgical intervention would be contemplated, but are more commonly associated with the advanced stages of cancer. However, a significant number of patients in the early stages of the disease, experience a suppression of the action of the hormone erythropoietin (EPO) on the bone marrow presumed to be caused by release of cytokines by the tumour itself [2]. The ensuing anaemia is usually mild and often asymptomatic. This has been termed cancer related anaemia. As well as suppression of the effect of EPO on the bone marrow, an inadequate EPO response to anaemia been shown to be a factor in the genesis of anaemia in malignancy [3—5]. Despite this, there is a quantitative rise in EPO concentration early in the disease making it potentially a sensitive marker of early anaemia development. Knowledge of prognostic factors is necessary in planning the most appropriate treatment regimens. Pre-operative haemoglobin concentration of less than 10 g/dl has been shown to have prognostic significance in patients with surgically resected stage I and II NSCLC [6]. Despite the much investigated role of EPO in the development of anaemia, no published study exists that investigates the relationship between pre-operative EPO concentration and prognosis in patients having surgical resection of primary lung cancer. This study evaluates the relationship between pre-operative EPO concentration, haemoglobin, and survival in patients having surgery for NSCLC.
2. Patients and methods Forty two patients were included having been referred for planned surgical resection of NSCLC between April 1998 and January 1999. All patients had 10 ml of blood drawn pre-operatively. Full blood
picture analysis was carried out on 4 ml in EDTA (Fysnex SE 9500, Japan), the remaining blood was spun down and the plasma stored at −80 ◦ C. None of the 42 patients had another documented cause for anaemia. Survival time was defined as the interval between the date of surgery to death, or the last day of follow-up in surviving patients. One hundred per cent follow-up was obtained with a median of 4.6 years with a range from 3.4 to 5.2 years in surviving patients. Pre-operative plasma samples at −80 ◦ C were thawed and immediate analysis, in the same batch, was performed using standard commercial ELISA assays for erythropoietin. (QuantikineR IVDR Lot No. 218078). The normal range on the assay was 2.5—10.5 mIU/ml. Patients were defined as anaemic if the haemoglobin concentration was less than 13.0 g/dl in males and less than 11.5 g/dl in females. Statistical methods: Spearman’s rank correlation was used to assess the relationship between EPO and both haemoglobin and PCV. Kruskal—Wallis ANOVA test was used to assess the difference in EPO concentration between patients with differing stage of disease. Fisher’s exact test was used to assess the relationship between resectability of the tumour and, EPO and haemoglobin, respectively. Kaplan—Meier survival curves were constructed as shown in the figures and the log-rank test statistic used to assess statistical significance. Multivariate analysis of survival was carried out using Cox’s proportional hazards test. Statistical significance was accepted when p < 0.05. Full ethical approval was obtained from the local Research Ethics Committee (App No. 397/03).
3. Results 3.1. Pre-operative characteristics Forty-two patients had surgery for NSCLC, seven female, 35 male. The median age was 68.1 years (range 42.6—77.8). Table 1 illustrates the preoperative variables measured. There was a negative correlation between EPO and both haemoglobin and PCV. (Spearman’s rank r = −0.353, p = 0.02 and r = −0.364, p = 0.018, respectively). This is entirely expected given the known physiological relationship between these variables. Of note, 17 patients (40.5%) had elevated pre-operative EPO and 15
Pre-operative plasma erythropoietin concentration and survival
331
Table 1 Pre-operative haematological variables in 42 non-small cell lung cancer patients Variable
Range
Median
EPO (mIU/ml) Haemoglobin (g/dl) Packed cell volume (l/l) Mean cell volume (fl)
3.7—56.4 8.5—16.8 0.264—0.506 75.1—105.8
9.4 13.3 0.401 92.8
(35.7%) were anaemic. Of the 17 with elevated preoperative EPO, only nine (52.9%) were anaemic.
3.2. Clinical and pathological staging All 42 patients were clinically staged as having resectable disease (cT1-3, cN0-1). All patients had chest X-ray, CT scan of chest and upper abdomen, and rigid bronchoscopy as minimum staging investigations. Other investigations such as CT brain, isotope bone scan were only carried out if clinically indicated. Patients with mediastinal lymphadenopathy greater than 10 mm in the short axis had mediastinoscopy. Pathological stage of tumour from TNM and stage (I—IV) was determined in each case after thoracotomy and histological assessment of the resected specimen. Of the 42 patients, 10 had irresectable disease, eight exploratory thoracotomy (19%) and two were turned down for surgery with N2 disease following mediastinoscopy. Seven of the eight patients who had exploratory thoracotomy were found to have stage IIIB disease. The other patient became unstable on clamping the pulmonary artery and so was unable to tolerate the level of resection required. Data on tumour stage are shown in Table 2. The other 10 patients with stage N2 (IIIA) disease were clinically staged as N01 but pathological staging revealed N2 disease after lymph node clearance at thoracotomy. Differences in pre-operative EPO concentration in patients with differing stage of disease were assessed using Kruskal—Wallis ANOVA test. This showed a significant difference in median EPO concentration between the different pathological
Table 2
Fig. 1 Scatterplot showing relationship between EPO and haemoglobin.
stages (p = 0.01). Similar analysis for haemoglobin showed no significant difference between the stages (p = 0.171) Seven of the seventeen patients (41.2%) with an elevated pre-operative EPO subsequently had irresectable disease, compared to only three of 25 (12.0%) with normal pre-operative EPO subsequently having irresectable disease. This relationship is statistically significant. (Fisher’s exact test p = 0.036). Five of the 15 anaemic patients (33.3%) had irresectable disease compared to five of 27 (18.5%) non-anaemic patients having irresectable disease. Interestingly, this is not a significant association (p = 0.239, Fisher’s exact test (Fig. 1)).
3.3. Survival following surgery Kaplan—Meier survival curves were drawn for pre-operative EPO and haemoglobin comparing normal and abnormal values. In patients with an elevated pre-operative EPO concentration (above 10.5 mIU/ml) survival was significantly less than those patients with a normal pre-operative EPO (log-rank; p = 0.026) (Fig. 2). Patients who were anaemic pre-operatively also appear to have reduced survival than those with normal haemoglobin, although statistical significance is not reached (log-rank; p = 0.074) (Fig. 3). As expected,
Post-operative pathological stage of tumours in 42 patients having surgery for NSCLC
Stage
Number
Percentage
Median EPO (mIU/ml)
Median haemoglobin (g/dl)
Ia IIa IIIaa IIIba
16 7 12 7
38.1 16.7 28.6 16.7
6.4* 10.0* 12.15* 13.9*
13.8** 13.4** 13.1** 12.2*
a * **
Stage I + II are resectable, III represents locally advanced disease, IIIa being resectable IIIb irresectable. p = 0.001 Kruskal—Wallis ANOVA. p = 0.171 Kruskal—Wallis ANOVA.
332
Fig. 2 Kaplan—Meier survival curve for patients with elevated vs. normal EPO level pre-operatively.
Fig. 3 Kaplan—Meier survival curve for anaemic patients vs. normal haemoglobin pre-operatively.
resection of the tumour had a significant influence on survival (Fig. 4). As we have shown, successful resection has a significant influence on survival, survival analyses for pre-operative EPO and haemoglobin were repeated on the sub group of patients who were able to have surgical resection of their tumours. Similar relationships are observed in the Kaplan—Meier curves although statistical significance was not achieved
I. Paul et al.
Fig. 5 Kaplan—Meier survival curve for patients with elevated vs. normal EPO level pre-operatively, excluding patients not resected.
Fig. 6 Kaplan—Meier survival curve for anaemic patients vs. normal haemoglobin pre-operatively, excluding patients not resected.
for EPO (log-rank; p = 0.172) or haemoglobin (logrank; p = 0.158) (Figs. 5 and 6). Multivariate analysis with Cox’s proportional hazards test using the pre-operative blood indices suggests EPO (ˇ = 0.057, p = 0.009) and not haemoglobin (ˇ = −0.101, p = 0.834) as an independent predictor of survival.
4. Discussion
Fig. 4 Kaplan—Meier survival curve for patients with resectable and irresectable disease.
Our study is the first of its kind examining the relationship between pre-operative EPO and long-term survival in patients having surgery for NSCLC. We recognise this study is on a relatively small number of patients (42), however the findings we present are significant. Looking at all patients in the study sample, raised plasma EPO pre-operatively has a significant bearing on subsequent survival postoperatively. The difference in survival between anaemic and non-anaemic patients pre-operatively was not statistically significant, perhaps due to the small sample size. A power calculation could not
Pre-operative plasma erythropoietin concentration and survival be carried out as there are no previous studies on EPO and survival following surgery for NSCLC. Preoperative haemoglobin concentration of less than 10 g/dl has been shown to have prognostic significance in patients with surgically resected stage I and II NSCLC [6]. In our study, however, only one patient (2.4%) fell into this category, and included patients with stage IIIa disease, therefore this data was felt not to be a comparable sample for the purposes of power calculation. It is interesting to note that that a high pre-operative EPO predicted a reduced survival but the presence of mild anaemia did not. Indeed multivariate analysis would suggest that pre-operative EPO is the only independent predictor of outcome among the haematological indices measured, although we recognise this can be hard to interpret with small numbers as in our case. The results demonstrate a significant difference with more advanced stages being associated with higher pre-operative EPO concentration. Pathological stage could be considered as a confounding factor accounting for the worse prognosis associated with raised pre-operative EPO. The data in Table 2 shows that patients who subsequently have stage IIIa and IIIb are the only groups with a median EPO concentration above the normal range. Larger studies are required to evaluate preoperative EPO and subsequent prognosis in patients with the same stage of disease. Rising EPO concentration with more advanced disease cannot be attributed to increasing anaemia in these groups. Table 2 shows that the group with stage IIIb disease are the only group with the median haemoglobin below the normal range and there is no significant difference in median haemoglobin concentration between the different pathological stages. Across the whole sample, eight of the 17 (47.1%) patients with elevated pre-operative EPO had a haemoglobin within the normal range. Evidence in the literature suggests cancer related anaemia is due to lack of an appropriate EPO response to a falling haemoglobin as a factor in the genesis of the anaemia [3—5]. These findings suggest there may be another, as yet unknown, relationship between NSCLC and EPO resulting in patients with a normal haemoglobin having elevated EPO concentration. The relationship between pre-operative EPO and subsequent resectability reflects the association between EPO and pathological stage mentioned previously. Our results suggest that an elevated preoperative EPO is significant not only in long-term prognosis but also in determining the subsequent resectability of the tumour. The precise pathological stage of NSCLC cannot be determined until after
333
operation and full analysis of the resected specimen is complete. As highlighted in our study, clinical staging is not always accurate and tumours can be upstaged or down staged at thoracotomy. In our study group all 42 patients were clinically staged as having resectable disease yet eight had irresectable disease at thoracotomy. This equates to an exploratory thoracotomy rate of 19% for this sample. Current guidelines from the British Thoracic Society in 2001 [9] state that exploratory thoracotomy rates should be between 5 and 10%. The rate for this sample is high (19%). It is however a small sample and other published series of similar size have shown equally high exploratory thoracotomy rates [7,8]. The exploratory thoracotomy rate in another series of 89 patients from our unit in a NSCLC study between September 2000 and January 2003 was only 3.9% [10]. Ultimately, the stage of the disease can only be determined pathologically after thoracotomy so no pre-operative staging investigation is 100% sensitive and specific. This highlights the importance of maximising the amount of data available in decision-making preoperatively. As there appears to be a relationship between EPO and resectability, the survival analyses were repeated in the subgroup of patients where resection was achieved. Similar trends exist although statistical significance is not achieved for either EPO or haemoglobin as this group only comprised 32 patients. Further larger studies are required to validate these findings and further investigate the biological relationship between EPO and NSCLC.
5. Conclusion This is a novel study into EPO and haemoglobin in the surgical patient with lung cancer. Raised pre-operative EPO and/or low haemoglobin, appear to be associated with reduced survival in these patients. On multivariate analysis, EPO appears the more significant of these variables. These indices are potentially of value in the pre-operative assessment of patients undergoing surgery for NSCLC. Larger prospective studies are required to validate these findings and to examine the biological relationship between non-small cell lung cancer and erythropoietin.
Acknowledgements The authors are grateful to Geraldine Savage and Sally Hagan for their help with erythropoietin assays.
334 The authors would like to acknowledge the support of the Royal Victoria Hospital, Belfast by means of a Royal Victoria Hospital Fellowship. Grant support from the Roy Castle Lung Cancer Foundation is also gratefully acknowledged.
References [1] Beguin Y. Erythropoietin and the anemia of cancer. Acta Clin Belg 1996;51-1:36—52. [2] Erslev AJ. Erythropoietin and anemia of cancer. Eur J Haematol 2000;64:353—8. [3] Cox R, Musial T, Gyde OHB. Reduced erythropoietin levels as a cause of anaemia in patients with lung cancer. Eur J Cancer Clin Oncol 1986;22(No. 4):511—4. [4] Ozguroglu M, Arun B, Demir G, Demirelli F, Mandel NM, Buyukunal E, Serdengecti S, Berkarda B. Serum erythropoietin level in anaemic cancer patients. Medical Oncol 2000;17:29—34.
I. Paul et al. [5] Miller CB, Jones RJ, Piantadosi S, Abeloff MD, Spivak JL. Decreased erythropoietin response in patients with the anaemia of cancer. N Eng J Med 1990;322:1689— 92. [6] Jazieh AR, Hussain M, Howington JA, Spencer HJ, Husain M, Grismer JT, Read RC. Prognostic factors in patients with surgically resected stages I and II non-small cell lung cancer. Ann Thorac Surg Oct 2000;70(4):1168—71. [7] Steinbaum SS, Uretzky ID, McAdams HP, Torrington KG, Cohen AJ. Exploratory thoracotomy for non-resectable lung cancer. Chest 1995;107:1058—61. [8] Morand G, Roeslin N. Exploratory thoracotomy of necessity in surgery of bronchial cancer. Ann Chir 1990;44(2): 133—7. [9] British Thoracic Society Guidelines. Guidelines on the selection of patients with lung cancer for surgery Thorax 2001; 56 89—108. [10] Jones JM, McGonigle N, McAnespie M, Graham AN, editors. Plasma, fibrinogen and serum C-reactive protein concentration predict resectability of non-small cell lung cancer. In: Presented at Society of Cardiothoracic Surgeons of Great Britain and Ireland, March 2004.
Pre-operative plasma erythropoietin concentration and survival following surgery for non-small cell lung cancer I. Paul a,∗, T.R.J. Lappin b, P. Maxwell c, A.N.J. Graham a a
Department of Cardiothoracic Surgery, Royal Victoria Hospital, Northern Ireland, Belfast, UK Department of Haematology, Belfast City Hospital, Northern Ireland, Belfast, UK c Department of Pathology, Queen’s University Belfast, Institute of Clinical Science, The Royal Group of Hospitals, Northern Ireland, Belfast, UK b
Received 10 June 2005 ; received in revised form 10 October 2005; accepted 31 October 2005 KEYWORDS Non-small cell lung carcinoma; Erythropoietin; Prognosis; Stage; Resectability; Haemoglobin
Summary Background: Suppression of the effect of the hormone erythropoietin (EPO) on the bone marrow, and an inadequate EPO response to anaemia have been shown to be factors in the genesis of cancer related anaemia. Low haemoglobin (Hb) concentration pre-operatively has been shown to have prognostic significance in patients with surgically resected NSCLC. This study investigates the relationship between pre-operative EPO and survival in patients having surgery for NSCLC. Methods: Pre-operative plasma EPO concentration and haemoglobin concentration were analysed in patients undergoing surgery for NSCLC between April 1998 and January 1999. Full follow-up was available for all patients. Results: Forty two patients were included. Median EPO concentration was 9.4 mIU/ml, range (3.7—56.4) with 17 patients (40.4%) having values above the normal range. Median haemoglobin concentration was 13.3 g/dl (range 8.5—16.8) with 15 patients (26%) anaemic pre-operatively. Pathological staging revealed 17 (40.4%) patients with stage I, 6 (14.3%) with stage II, 19 (45.3%) with stage III disease. Ten patients had irresectable disease. There was a significant difference in median EPO but not haemoglobin concentration, between the different pathological stages. Survival was significantly lower in patients with pre-operative EPO >10.5 mIU. Conclusions: Raised pre-operative EPO is associated with reduced survival in patients having surgery for NSCLC. Its measurement should be considered in the pre-operative
∗
Corresponding author at: Department of Cardiothoracic Surgery, Royal Victoria Hospital, Grosvenor Road, Northern Ireland, Belfast BT12 6BA, UK. Tel.: +44 28 9024 0503; fax: +44 28 9024 0503. E-mail address: [email protected] (I. Paul). 0169-5002/$ — see front matter © 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.lungcan.2005.10.020
330
I. Paul et al. assessment of patients undergoing surgery for NSCLC. Further research is required to further investigate the biological relationship between EPO and NSCLC. © 2005 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Anaemia is a common finding in patients with solid tumours, including non-small cell lung carcinoma (NSCLC). Factors involved in the pathogenesis of the anaemia of cancer are multiple. Haemodilution, bleeding from tumour erosion, hypersplenism, haemolysis, nutritional deficiencies, bone marrow infiltration, chemotherapy and radiotherapy have all been highlighted as contributing factors [1]. Many of these causes usually do not develop early in the course of the disease, when surgical intervention would be contemplated, but are more commonly associated with the advanced stages of cancer. However, a significant number of patients in the early stages of the disease, experience a suppression of the action of the hormone erythropoietin (EPO) on the bone marrow presumed to be caused by release of cytokines by the tumour itself [2]. The ensuing anaemia is usually mild and often asymptomatic. This has been termed cancer related anaemia. As well as suppression of the effect of EPO on the bone marrow, an inadequate EPO response to anaemia been shown to be a factor in the genesis of anaemia in malignancy [3—5]. Despite this, there is a quantitative rise in EPO concentration early in the disease making it potentially a sensitive marker of early anaemia development. Knowledge of prognostic factors is necessary in planning the most appropriate treatment regimens. Pre-operative haemoglobin concentration of less than 10 g/dl has been shown to have prognostic significance in patients with surgically resected stage I and II NSCLC [6]. Despite the much investigated role of EPO in the development of anaemia, no published study exists that investigates the relationship between pre-operative EPO concentration and prognosis in patients having surgical resection of primary lung cancer. This study evaluates the relationship between pre-operative EPO concentration, haemoglobin, and survival in patients having surgery for NSCLC.
2. Patients and methods Forty two patients were included having been referred for planned surgical resection of NSCLC between April 1998 and January 1999. All patients had 10 ml of blood drawn pre-operatively. Full blood
picture analysis was carried out on 4 ml in EDTA (Fysnex SE 9500, Japan), the remaining blood was spun down and the plasma stored at −80 ◦ C. None of the 42 patients had another documented cause for anaemia. Survival time was defined as the interval between the date of surgery to death, or the last day of follow-up in surviving patients. One hundred per cent follow-up was obtained with a median of 4.6 years with a range from 3.4 to 5.2 years in surviving patients. Pre-operative plasma samples at −80 ◦ C were thawed and immediate analysis, in the same batch, was performed using standard commercial ELISA assays for erythropoietin. (QuantikineR IVDR Lot No. 218078). The normal range on the assay was 2.5—10.5 mIU/ml. Patients were defined as anaemic if the haemoglobin concentration was less than 13.0 g/dl in males and less than 11.5 g/dl in females. Statistical methods: Spearman’s rank correlation was used to assess the relationship between EPO and both haemoglobin and PCV. Kruskal—Wallis ANOVA test was used to assess the difference in EPO concentration between patients with differing stage of disease. Fisher’s exact test was used to assess the relationship between resectability of the tumour and, EPO and haemoglobin, respectively. Kaplan—Meier survival curves were constructed as shown in the figures and the log-rank test statistic used to assess statistical significance. Multivariate analysis of survival was carried out using Cox’s proportional hazards test. Statistical significance was accepted when p < 0.05. Full ethical approval was obtained from the local Research Ethics Committee (App No. 397/03).
3. Results 3.1. Pre-operative characteristics Forty-two patients had surgery for NSCLC, seven female, 35 male. The median age was 68.1 years (range 42.6—77.8). Table 1 illustrates the preoperative variables measured. There was a negative correlation between EPO and both haemoglobin and PCV. (Spearman’s rank r = −0.353, p = 0.02 and r = −0.364, p = 0.018, respectively). This is entirely expected given the known physiological relationship between these variables. Of note, 17 patients (40.5%) had elevated pre-operative EPO and 15
Pre-operative plasma erythropoietin concentration and survival
331
Table 1 Pre-operative haematological variables in 42 non-small cell lung cancer patients Variable
Range
Median
EPO (mIU/ml) Haemoglobin (g/dl) Packed cell volume (l/l) Mean cell volume (fl)
3.7—56.4 8.5—16.8 0.264—0.506 75.1—105.8
9.4 13.3 0.401 92.8
(35.7%) were anaemic. Of the 17 with elevated preoperative EPO, only nine (52.9%) were anaemic.
3.2. Clinical and pathological staging All 42 patients were clinically staged as having resectable disease (cT1-3, cN0-1). All patients had chest X-ray, CT scan of chest and upper abdomen, and rigid bronchoscopy as minimum staging investigations. Other investigations such as CT brain, isotope bone scan were only carried out if clinically indicated. Patients with mediastinal lymphadenopathy greater than 10 mm in the short axis had mediastinoscopy. Pathological stage of tumour from TNM and stage (I—IV) was determined in each case after thoracotomy and histological assessment of the resected specimen. Of the 42 patients, 10 had irresectable disease, eight exploratory thoracotomy (19%) and two were turned down for surgery with N2 disease following mediastinoscopy. Seven of the eight patients who had exploratory thoracotomy were found to have stage IIIB disease. The other patient became unstable on clamping the pulmonary artery and so was unable to tolerate the level of resection required. Data on tumour stage are shown in Table 2. The other 10 patients with stage N2 (IIIA) disease were clinically staged as N01 but pathological staging revealed N2 disease after lymph node clearance at thoracotomy. Differences in pre-operative EPO concentration in patients with differing stage of disease were assessed using Kruskal—Wallis ANOVA test. This showed a significant difference in median EPO concentration between the different pathological
Table 2
Fig. 1 Scatterplot showing relationship between EPO and haemoglobin.
stages (p = 0.01). Similar analysis for haemoglobin showed no significant difference between the stages (p = 0.171) Seven of the seventeen patients (41.2%) with an elevated pre-operative EPO subsequently had irresectable disease, compared to only three of 25 (12.0%) with normal pre-operative EPO subsequently having irresectable disease. This relationship is statistically significant. (Fisher’s exact test p = 0.036). Five of the 15 anaemic patients (33.3%) had irresectable disease compared to five of 27 (18.5%) non-anaemic patients having irresectable disease. Interestingly, this is not a significant association (p = 0.239, Fisher’s exact test (Fig. 1)).
3.3. Survival following surgery Kaplan—Meier survival curves were drawn for pre-operative EPO and haemoglobin comparing normal and abnormal values. In patients with an elevated pre-operative EPO concentration (above 10.5 mIU/ml) survival was significantly less than those patients with a normal pre-operative EPO (log-rank; p = 0.026) (Fig. 2). Patients who were anaemic pre-operatively also appear to have reduced survival than those with normal haemoglobin, although statistical significance is not reached (log-rank; p = 0.074) (Fig. 3). As expected,
Post-operative pathological stage of tumours in 42 patients having surgery for NSCLC
Stage
Number
Percentage
Median EPO (mIU/ml)
Median haemoglobin (g/dl)
Ia IIa IIIaa IIIba
16 7 12 7
38.1 16.7 28.6 16.7
6.4* 10.0* 12.15* 13.9*
13.8** 13.4** 13.1** 12.2*
a * **
Stage I + II are resectable, III represents locally advanced disease, IIIa being resectable IIIb irresectable. p = 0.001 Kruskal—Wallis ANOVA. p = 0.171 Kruskal—Wallis ANOVA.
332
Fig. 2 Kaplan—Meier survival curve for patients with elevated vs. normal EPO level pre-operatively.
Fig. 3 Kaplan—Meier survival curve for anaemic patients vs. normal haemoglobin pre-operatively.
resection of the tumour had a significant influence on survival (Fig. 4). As we have shown, successful resection has a significant influence on survival, survival analyses for pre-operative EPO and haemoglobin were repeated on the sub group of patients who were able to have surgical resection of their tumours. Similar relationships are observed in the Kaplan—Meier curves although statistical significance was not achieved
I. Paul et al.
Fig. 5 Kaplan—Meier survival curve for patients with elevated vs. normal EPO level pre-operatively, excluding patients not resected.
Fig. 6 Kaplan—Meier survival curve for anaemic patients vs. normal haemoglobin pre-operatively, excluding patients not resected.
for EPO (log-rank; p = 0.172) or haemoglobin (logrank; p = 0.158) (Figs. 5 and 6). Multivariate analysis with Cox’s proportional hazards test using the pre-operative blood indices suggests EPO (ˇ = 0.057, p = 0.009) and not haemoglobin (ˇ = −0.101, p = 0.834) as an independent predictor of survival.
4. Discussion
Fig. 4 Kaplan—Meier survival curve for patients with resectable and irresectable disease.
Our study is the first of its kind examining the relationship between pre-operative EPO and long-term survival in patients having surgery for NSCLC. We recognise this study is on a relatively small number of patients (42), however the findings we present are significant. Looking at all patients in the study sample, raised plasma EPO pre-operatively has a significant bearing on subsequent survival postoperatively. The difference in survival between anaemic and non-anaemic patients pre-operatively was not statistically significant, perhaps due to the small sample size. A power calculation could not
Pre-operative plasma erythropoietin concentration and survival be carried out as there are no previous studies on EPO and survival following surgery for NSCLC. Preoperative haemoglobin concentration of less than 10 g/dl has been shown to have prognostic significance in patients with surgically resected stage I and II NSCLC [6]. In our study, however, only one patient (2.4%) fell into this category, and included patients with stage IIIa disease, therefore this data was felt not to be a comparable sample for the purposes of power calculation. It is interesting to note that that a high pre-operative EPO predicted a reduced survival but the presence of mild anaemia did not. Indeed multivariate analysis would suggest that pre-operative EPO is the only independent predictor of outcome among the haematological indices measured, although we recognise this can be hard to interpret with small numbers as in our case. The results demonstrate a significant difference with more advanced stages being associated with higher pre-operative EPO concentration. Pathological stage could be considered as a confounding factor accounting for the worse prognosis associated with raised pre-operative EPO. The data in Table 2 shows that patients who subsequently have stage IIIa and IIIb are the only groups with a median EPO concentration above the normal range. Larger studies are required to evaluate preoperative EPO and subsequent prognosis in patients with the same stage of disease. Rising EPO concentration with more advanced disease cannot be attributed to increasing anaemia in these groups. Table 2 shows that the group with stage IIIb disease are the only group with the median haemoglobin below the normal range and there is no significant difference in median haemoglobin concentration between the different pathological stages. Across the whole sample, eight of the 17 (47.1%) patients with elevated pre-operative EPO had a haemoglobin within the normal range. Evidence in the literature suggests cancer related anaemia is due to lack of an appropriate EPO response to a falling haemoglobin as a factor in the genesis of the anaemia [3—5]. These findings suggest there may be another, as yet unknown, relationship between NSCLC and EPO resulting in patients with a normal haemoglobin having elevated EPO concentration. The relationship between pre-operative EPO and subsequent resectability reflects the association between EPO and pathological stage mentioned previously. Our results suggest that an elevated preoperative EPO is significant not only in long-term prognosis but also in determining the subsequent resectability of the tumour. The precise pathological stage of NSCLC cannot be determined until after
333
operation and full analysis of the resected specimen is complete. As highlighted in our study, clinical staging is not always accurate and tumours can be upstaged or down staged at thoracotomy. In our study group all 42 patients were clinically staged as having resectable disease yet eight had irresectable disease at thoracotomy. This equates to an exploratory thoracotomy rate of 19% for this sample. Current guidelines from the British Thoracic Society in 2001 [9] state that exploratory thoracotomy rates should be between 5 and 10%. The rate for this sample is high (19%). It is however a small sample and other published series of similar size have shown equally high exploratory thoracotomy rates [7,8]. The exploratory thoracotomy rate in another series of 89 patients from our unit in a NSCLC study between September 2000 and January 2003 was only 3.9% [10]. Ultimately, the stage of the disease can only be determined pathologically after thoracotomy so no pre-operative staging investigation is 100% sensitive and specific. This highlights the importance of maximising the amount of data available in decision-making preoperatively. As there appears to be a relationship between EPO and resectability, the survival analyses were repeated in the subgroup of patients where resection was achieved. Similar trends exist although statistical significance is not achieved for either EPO or haemoglobin as this group only comprised 32 patients. Further larger studies are required to validate these findings and further investigate the biological relationship between EPO and NSCLC.
5. Conclusion This is a novel study into EPO and haemoglobin in the surgical patient with lung cancer. Raised pre-operative EPO and/or low haemoglobin, appear to be associated with reduced survival in these patients. On multivariate analysis, EPO appears the more significant of these variables. These indices are potentially of value in the pre-operative assessment of patients undergoing surgery for NSCLC. Larger prospective studies are required to validate these findings and to examine the biological relationship between non-small cell lung cancer and erythropoietin.
Acknowledgements The authors are grateful to Geraldine Savage and Sally Hagan for their help with erythropoietin assays.
334 The authors would like to acknowledge the support of the Royal Victoria Hospital, Belfast by means of a Royal Victoria Hospital Fellowship. Grant support from the Roy Castle Lung Cancer Foundation is also gratefully acknowledged.
References [1] Beguin Y. Erythropoietin and the anemia of cancer. Acta Clin Belg 1996;51-1:36—52. [2] Erslev AJ. Erythropoietin and anemia of cancer. Eur J Haematol 2000;64:353—8. [3] Cox R, Musial T, Gyde OHB. Reduced erythropoietin levels as a cause of anaemia in patients with lung cancer. Eur J Cancer Clin Oncol 1986;22(No. 4):511—4. [4] Ozguroglu M, Arun B, Demir G, Demirelli F, Mandel NM, Buyukunal E, Serdengecti S, Berkarda B. Serum erythropoietin level in anaemic cancer patients. Medical Oncol 2000;17:29—34.
I. Paul et al. [5] Miller CB, Jones RJ, Piantadosi S, Abeloff MD, Spivak JL. Decreased erythropoietin response in patients with the anaemia of cancer. N Eng J Med 1990;322:1689— 92. [6] Jazieh AR, Hussain M, Howington JA, Spencer HJ, Husain M, Grismer JT, Read RC. Prognostic factors in patients with surgically resected stages I and II non-small cell lung cancer. Ann Thorac Surg Oct 2000;70(4):1168—71. [7] Steinbaum SS, Uretzky ID, McAdams HP, Torrington KG, Cohen AJ. Exploratory thoracotomy for non-resectable lung cancer. Chest 1995;107:1058—61. [8] Morand G, Roeslin N. Exploratory thoracotomy of necessity in surgery of bronchial cancer. Ann Chir 1990;44(2): 133—7. [9] British Thoracic Society Guidelines. Guidelines on the selection of patients with lung cancer for surgery Thorax 2001; 56 89—108. [10] Jones JM, McGonigle N, McAnespie M, Graham AN, editors. Plasma, fibrinogen and serum C-reactive protein concentration predict resectability of non-small cell lung cancer. In: Presented at Society of Cardiothoracic Surgeons of Great Britain and Ireland, March 2004.