- Email: [email protected]
A study of lymph node ratio as a prognostic marker in colon cancer
Available online at www.sciencedirect.com
EJSO 34 (2008) 771e775
www.ejso.com
A study of lymph node ratio as a prognostic marker in colon cancer K. Derwinger*, G. Carlsson, B. Gustavsson ¨ stra, 41685 Gothenburg, Sweden Department of Surgery, Sahlgrenska University Hospital/O Accepted 1 November 2007 Available online 20 February 2008
Abstract Aim: The aim of this study was to evaluate and describe the lymph node ratio (LNR) as a prognostic parameter for patients with colon cancer. As lymphatic involvement is the key, focus was set at stage III disease. Interest was directed at the possibility of identifying high-risk groups and the clinical implementation and consequence. Method: The study was retrospective using a database of clinical data of all cancer patients treated at our unit. It has been continuous in registration, inclusion and update since 1999 including survival and clinical features. All patients (n ¼ 265) diagnosed with stage III colon cancer during 1999e2003 were included for the study. LNR was calculated and quartile groups were created. LNR and associated parameters were analysed towards 3-year disease-free survival (DFS). Basic patient data as well as surgery, pathology and postoperative treatment were taken into consideration. Results: Significant differences in disease-free survival were found for TNM N-status, tumour differentiation grade and LNR quartile group. There was a difference in 3-year DFS from 80% in LNR group 1 compared with less than 30% in group 4. These results were of prognostic interest both independently and in interaction with each other. High-risk groups could be identified and in the worst prognosis LNR group we also found a tendency towards more side effects with adjuvant chemotherapy. Conclusion: The lymph node ratio, the quota between the number of lymph node metastasis and assessed lymph nodes, is a highly significant ( p < 0.001) prognostic factor in stage III colon cancer. It can be an aid in identifying risk groups that could benefit from a more intense postoperative surveillance and possibly bring changes in adjuvant treatment strategy. More studies of clinical data, genetic and biochemical markers are needed in this patient group to understand the possible difference in tumour behaviour and tailor the treatment. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Colon cancer; Prognosis; Survival; Lymph node ratio
Introduction Colon cancer is one of the most common cancer forms in Sweden as well as in Western Europe and the incidence is slowly increasing.1 In being a common cancer form all information and knowledge that can assist in optimizing the process of staging, prognostics and choice of treatment strategy are of importance. The most important prognostic factor that influences long-term survival is the knowledge of the presence of lymph node involvement or distant metastases. Lymph node involvement is the main factor for including patients in adjuvant chemotherapeutic protocols.2 Failure to make an adequate staging assessment might deprive a patient of beneficial adjuvant therapy with resulting shorter survival. Simultaneously, the risk of over treating * Corresponding author. Tel.: þ46313434219; fax: þ46313435930. E-mail address: [email protected] (K. Derwinger). 0748-7983/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.ejso.2007.11.002
should also be considered with unnecessary risk of adverse effects for the patients. The most commonly used system for cancer staging is the TNM classification. It takes the main parameters of local growth, regional lymph node involvement and presence of distant spread into consideration.3 It is systematic and functional but within stage III disease there are prognostic limitations. One of these is that stage III is often treated as an entity and that the differences in prognosis within the group are not clear. This does affect the treatment and the follow-up and limits the possibility of a tailored strategy. The importance of the pathologist’s work and adherence to quality standards is also known.4,5 Another developing and interesting field that is rapidly expanding is that of biomolecular and genetic markers that might give information for prognosis and as aid in choosing the line of therapy.6 Many new techniques are also developed for metastasis detection.7 This will have an impact on cancer staging and
K. Derwinger et al. / EJSO 34 (2008) 771e775
772
primarily in differing between stage I/II and stage III and in the end on stage-specific survival. Node related factors, other than present ones, that can influence survival and thus can be of prognostic interest include lymph node size and distribution.8,9 Another possibility is the lymph node ratio (LNR) which is the quota between positive and assessed lymph node. This parameter is well described as a significant predictor in gastric cancer as well as in breast cancer and there is evidence of usefulness in colorectal cancer.10e12 The aim of this study was to evaluate and describe the lymph node ratio as a prognostic parameter for patients with colon cancer. As lymphatic involvement is the key, focus was set at stage III disease. Interest was directed at the possibility of identifying high-risk groups and the clinical implementation and consequence. The study is a retrospective analysis of data from a high volume surgical unit in a university hospital.
as this outcome parameter as it is an accepted way to reflect on the future long-term prognosis.14
Method
Results
Material and method
The patients and the surgery
¨ stra At the Department of Surgery, Sahlgrenska/O University Hospital, Gothenburg, we are continuously making a registration of detailed clinical and pathological data. The registration is consecutive since 1999 for all patients treated for colorectal cancer. Included in the database are also a continuous follow-up regarding treatment, recurrence and survival. In staging we use the TNM classification, version 5, 1997.13 During the period 1999e2003, 718 patients were diagnosed with colon cancer at our unit. For this study, we selected all patients (n ¼ 265) with stage III disease. The overall stage distribution during the period was stage (I) 9%, (II) 37%, (III) 36% and stage (IV) 18%. All the patients were treated following the same guidelines including postoperative surveillance. We studied basic clinical parameters such as sex, age, diagnosis, cancer location, operation as well as pathology data such as lymph nodes and differentiation grade. Interest was focused on number of nodes with metastasis growth correlated to the number of assessed nodes. Lymph node ratio was calculated as the quota between metastasis positive nodes and assessed lymph nodes. LNR groups were created on the basis of ratio quartiles. As prognostic instrument we used 3-year disease-free survival data
The average age in the material was 72 years. There was a slight female predominance. The preoperative staging at our unit was performed with chest X-ray and liver examination (CT or ultrasound) and was completed in 98% of elective patients. There was no difference in the spectra of surgical resections between the LNR groups with right hemicolectomy being the most common. Neither were there any differences in surgical panorama, age, sex or hospital stay when grouping either for LNR group, N-status or differentiation grade. The proportion of emergency procedures in the entire group was 24% with a variation from 20% in LNR group 1 to 29% in group 4. The median number of assessed nodes was 11 (range 4e160). The number of assessed nodes did not differ with any grouping but the number of metastatic nodes increased naturally with LNR group. Patient characteristics and pathology data details as presented in Table 1.
Statistical method We used the JMP 4/SAS software for statistical analysis. Basic patient demographic data were set by distribution statistics with ANOVA or contingency tables with Chi-square analysis. The KaplaneMeier method was used to calculate cumulative survival and Log rank test was used to compare survival differences between LNR groups. The same analyses were made for TNM N-status (node status: N1 or N2) as well as differentiation grade. We also performed a COX proportional hazard analysis to explore their significance and possible co-variation. All the analyses were also reexamined after exclusion of any patients in whom less than 12 nodes had been assessed, as by UICC guidelines, to ascertain that eventual results retained their significance.
Adjuvant therapy All stage III patients (n ¼ 265) were discussed in a multimodal treatment conference and if eligible were treated by
Table 1 Patients’ demographics and pathology data in LNR groups LNR group
LNR internal
1 2 3 4
0e0.125 0.126e0.266 0.267e0.450 0.451e1
Total
Number
63 71 66 65 265
Age, average (range)
Assessed nodes, medium (range)
Metastasis nodes, medium(range)
Differentiation grade (high/ medium/poor)
Adjuvant treatment (yes/no)
N-status (N1/N2)
70 72 71 71
13 12 10 11
1 2 3.5 6
5/45/13 3/48/20 1/34/31 0/31/34
39/24 44/27 36/30 38/27
63/0 53/18 32/34 8/57
8/159/98
157/108
156/109
(25e91) (44e90) (27e92) (32e90)
72 (25e92)
(8e160) (4e29) (5e31) (6e22)
11 (4e160)
(1e2) (1e5) (2e13) (3e20)
2 (1e20)
K. Derwinger et al. / EJSO 34 (2008) 771e775
our oncologists. There was no difference between the LNR groups in frequency of inclusion into adjuvant chemotherapy protocols (n ¼ 157). Neither was there any difference in the choice of regime. The most frequent protocol was with 5-FU and Leucovorin accounting for more than 95% of the treatments. There was a small tendency in LNR group 4 of having more side effects and thus more often than LNR groups 1e3 not completing the chemotherapeutic protocol. The difference was seen but does not reach significance ( p < 0.1). When examining the cohort that was not eligible for chemotherapy some data differed. As age is a limiting factor, the oldest patients are rarely eligible for adjuvant treatment. In this group (n ¼ 108) the mean age was 80 (range 41e92). There was no significant difference in the number of assessed nodes or tumour stages. Nor was there any significant difference in surgical panorama or hospital stay for the group. Age was the most frequent factor in this decision but concomitant disease also played a role. When separately examining the prognostic factors in this cohort LNR and N-status, although still significant, had less impact on cancer-specific survival whilst the differentiation grade was relatively more important ( p < 0.01). The nodes and differentiation The number of assessed positive nodes increased in LNR group to a significant level ( p < 0.05). There was no difference between any groupings in total number of assessed nodes. The analysis of survival and comparison between TNM system N1 (1e3 positive nodes, n ¼ 161) and N2 (more than three positive nodes, n ¼ 104) showed a significant difference in 3-year disease-free survival ( p < 0.001). There was also a co-variation between differentiation grade and node status both in prevalence and affecting survival. A worse differentiation grade was more common in N2-status than in N1. The aspect of tumour differentiation grade is also a significant ( p < 0.001) prognostic factor for survival where a worse (lower) differentiation grade corresponds to worse outcome. Between these groupings, there was no difference in analyses regarding sex, age or number of assessed nodes. When re-examining the same tests after exclusion of the cases where less than 12 nodes were assessed, we retained the same significance of our results. The differentiation grade also affected survival at a significant level ( p < 0.05) within each N group, giving an interesting co-variation.
773
Figure 1. Disease-free survival in LNR groups.
II levels. Though still in stage III, LNR group 4 in contrast had a far worse survival prognosis with a 29% 3-year DFS. A poor differentiation grade was more common in LNR group 4 whilst a high grade was uncommon in stage III cancer in our material. As LNR is a quota the N2 node status was more frequent in the higher LNR groups. In crossexamination of N-status (N1/N2) with differentiation we found significant ( p < 0.05) difference in survival in each category with the worst prognosis being N2 and poor differentiation grade. However, this did not fully match the worse prognosis of LNR group 4. In multivariate analysis LNR remained as a highly significant factor along with differentiation grade. Statistical analysis results are summarized in Table 2. Discussion Staging Staging is important in many aspects. It often decides the postoperative treatment and follow-up. Thus it affects the individual patient as the risk of harm from examination or drug side effect should be balanced by the risk of recurrence and chance of cure. It is also in some economic systems a needed aid in allocation of limited healthcare resources. Tools that can aid us in directing, individualizing and optimizing the postoperative process are of importance. The overall most significant prognostic indicator is the cancer stage either by Dukes or TNM stage IeIV.15 Absence of
LNR and survival The entire cohort (n ¼ 265) was divided into four groups after ratio quartiles. When comparing both overall survival and disease-free survival stratified by LNR group we found significant differences ( p < 0.001) (Fig. 1). In Cox proportional hazard analysis the risk ratio between LNR groups 1e2 was 1.8, LNR groups 2e3: 1.4 and groups 3e4: 2.0. LNR group 1 had a 3-year survival prognosis of 80% and thus well comparable to normally accepted stage
Table 2 Summary of statistical results Factor
Univariate, p-value
Multivariate (risk ratio (CI), p-value)
LNR N-status (N1/N2) Differentiation grade (medium/poor) Adjuvant therapy (yes/no)
<0.0001 <0.001 <0.001
10.6 (3.2e31.8), <0.0002 2.0 (1.1e3.9), <0.04 2.8 (1.2e6.1), <0.0006
<0.001
3.1 (2.1e4.7), <0.001
774
K. Derwinger et al. / EJSO 34 (2008) 771e775
lymph node involvement (N0 status) or distant spread (M0 status) means stage I or II disease and is thus strong positive prognostic markers. At the other end metastasis to other organs (M1-status), meaning stage IV disease, indicate a poor prognosis for long-term survival. In between lays stage III colon cancer which as shown is a heterogeneous group with a more variable outcome. Pathology A key factor in this process is the pathologist.16 It has been shown in several studies that their preparation, method and analysis will affect the report and thus also the clinical actions.17,18 The number of assessed lymph nodes needed for an adequate staging is set at 12 nodes in both national and international standard.19 This is a quality factor not to be neglected and a possible prognostic tool should be verified by data meeting this requirement as has been done here.20 Ratio calculation should be based on full node assessment but with focus on the nodes along the central spread vessels where a positive finding is more probable. This is often sub-specified in the reports from some pathology departments and might be of importance as the total number of assessed nodes is increasing. New possibilities of finding signs of cancer spread by PCR is also under development. It will be of interest though it will primarily increase the accuracy in defining between stage I/II and stage III disease. This in turn will affect the stage-specific survival prognostics through a possible stage migration. A ratio One possibility of further prognostic information within stage III disease is the lymph node ratio. It has a significant ( p < 0.001) impact in disease-free survival prognosis as shown in Fig. 1 and remained significant in a multivariate analysis. It can be an aid in finding risk groups where extra postoperative attention should be given. Possible factors contributing to the ratio prognostics are the number of assessed and positive nodes but also the differentiation grade as it covariates. With worse grade it is more frequent with more positive nodes and higher LNR group. The higher prevalence of emergency surgery in LNR group 4 can also contribute to worsening the outcome.21 However, separate analysis of the emergency cohort revealed the same ratio patterns and it could not by itself explain the prognostic difference. We believe that the ratio is a significant continuous variable in accordance with previous studies.18 Possibly it reflects tumour biology and the total cancer burden. Still, a grouping makes it more defined and facilitates the identification of risk groups. Other studies have tested quartile groupings with similar results and thus a possibility of comparison and confirmation.12,22 A higher number of groups make data more difficult to handle whilst fewer make risk categories difficult to distinguish.
Treatment and side effects The main line of chemotherapy for stage III disease is by 5-FU and Leucovorin, applying to more than 95% in this material.19 There was no difference in treatment between the ratio groups. Interesting factors that can alter or affect tumour malignancy potential and response to chemotherapy include genetics, bio-molecular markers and polymorphisms but these are under development and need further data and studies.23e25 Interesting is also the possible discussion of the treatment hazards and choice of therapy in a low-risk group. Whilst high-risk patients in group 4 might benefit from the most aggressive regimes, it is more difficult to advocate the same for low-risk patients in group 1. The development of tailored treatments might aid us in these questions. Regarding the cohort of patients that were not eligible for chemotherapy their overall prognosis was worse. This should be considered with the knowledge that age is a prognostic factor and that this group is older and more burdened by concomitant disease. Therefore, the clinical importance of the prognostic parameters and their significance is difficult to assess in this group. The study and beyond We believe that this material confirms the use of LNR as a possible prognostic tool. Our data and results are in line with and strengthen previous studies.12,22 In our opinion it does reflect the total tumour burden and biological properties. The study also confirms that the node related factors are and remain among the strongest prognostic parameters. The differentiation grade is of some importance and an interesting factor but in the literature its role is unclear. We were able to identify a high-risk patient group which could benefit from a more intense postoperative surveillance and possibly changes in the adjuvant treatment strategy. Additional studies must be made to find which drugs and regimes that can possibly give this group a better chance of long-term survival. One weakness in our material is the relatively small number of patients when compared to other studies. We believe this to be well compensated by the fact that the material is unselected and population based. We have good possibilities to survey and follow-up of the patients. The patients were treated at the same hospital mainly by the same staff and with same treatment strategies during this period. All were included, registered and treated using the same guidelines. Conclusion The lymph node ratio, the quota between the number of lymph node metastases and assessed lymph nodes, is a highly significant ( p < 0.001) prognostic factor in stage III colon cancer. It can be an aid in identifying risk groups that could benefit from a more intense postoperative surveillance and possibly changes in adjuvant treatment
K. Derwinger et al. / EJSO 34 (2008) 771e775
strategy. More studies of clinical data, genetic and biochemical markers are needed in this patient group to understand the possible difference in tumour behaviour and tailor the treatment. Conflict of interest The authors have no conflict of interest.
Acknowledgement We like to express our gratitude to the staff of our oncology laboratory unit assisting in the collection and registration of data and samples.
References 1. Oncological centre in Western Region of Sweden. Regional guidelines and statistics; 2005. 2. Glimelius B, Dahl O, Cedermark B, et al. Adjuvant chemotherapy in colorectal cancer: a joint analysis of randomised trials by the nordic gastrointestinal tumour adjuvant therapy group. Acta Oncol 2005; 44(8):904–12. 3. Kehoe J, Khatri VP. Staging and prognosis of colon cancer. Surg Oncol Clin N Am 2006;15(1):129–46. 4. Lemmens VE, Verheij CD, Janssen-Heijnen ML, et al. Mixed adherence to clinical practice guidelines for colorectal cancer in the southern Netherlands in 2002. Eur J Surg Oncol 2006;32(2):168–73. 5. Ratto C, Sofo L, Ippoliti M, et al. Accurate lymph-node detection in colorectal specimens resected for cancer is of prognostic significance. Dis Colon Rectum 1999;42(2):143–54. discussion 154e8. 6. Engwegen JY, Gast MC, Schellens JH, et al. Clinical proteomics: searching for better tumour markers with SELDI-TOF mass spectrometry. Trends Pharmacol Sci 2006;27(5):251–9. 7. Ohlsson L, Hammarstrom ML, Israelsson A, et al. Biomarker selection for detection of occult tumour cells in lymph nodes of colorectal cancer patients using real-time quantitative RT-PCR. Br J Cancer 2006; 95(2):218–25. 8. Kobayashi H, Ueno H, Hashiguchi Y, et al. Distribution of lymph node metastasis is a prognostic index in patients with stage III colon cancer. Surgery 2006;139(4):516–22.
775
9. Dhar DK, Yoshimura H, Kinukawa N, et al. Metastatic lymph node size and colorectal cancer prognosis. J Am Coll Surg 2005;200(1): 20–8. 10. Inoue K, Nakane Y, Iiyama H, et al. The superiority of ratio-based lymph node staging in gastric carcinoma. Ann Surg Oncol 2002; 9(1):27–34. 11. Voordeckers M, Vinh-Hung V, Van de Steene J, et al. The lymph node ratio as prognostic factor in node-positive breast cancer. Radiother Oncol 2004;70(3):225–30. 12. Berger AC, Sigurdson ER, LeVoyer T, et al. Colon cancer survival is associated with decreasing ratio of metastatic to examined lymph nodes. J Clin Oncol 2005;23(34):8706–12. 13. UICC/AJCC. Manual of TNM classification of malignant tumors. 5th ed.; 1997. 14. Sargent DJ, Wieand HS, Haller DG, et al. Disease-free survival versus overall survival as a primary end point for adjuvant colon cancer studies: individual patient data from 20,898 patients on 18 randomized trials. J Clin Oncol 2005;23(34):8664–70. 15. Wolters U, Stutzer H, Keller HW, et al. Colorectal cancer e a multivariate analysis of prognostic factors. Eur J Surg Oncol 1996;22(6):592–7. 16. Compton CC. Key issues in reporting common cancer specimens: problems in pathologic staging of colon cancer. Arch Pathol Lab Med 2006;130(3):318–24. 17. Lemmens VE, van Lijnschoten I, Janssen-Heijnen ML, et al. Pathology practice patterns affect lymph node evaluation and outcome of colon cancer: a population-based study. Ann Oncol 2006;17(12):1803–9. 18. Jestin P, Pahlman L, Glimelius B, et al. Cancer staging and survival in colon cancer is dependent on the quality of the pathologists’ specimen examination. Eur J Cancer 2005;41(14):2071–8. 19. National Board of Health in Sweden. National guidelines for colorectal cancer; 2006. 20. Derwinger K, Carlsson G, Gustavsson B. Stage migration in colorectal cancer related to improved lymph node assessment. Eur J Surg Oncol 2007 Sep;33(7):849–53. 21. Jestin P, Nilsson J, Heurgren M, et al. Emergency surgery for colonic cancer in a defined population. Br J Surg 2005;92(1):94–100. 22. Lee HY, Choi HJ, Park KJ, et al. Prognostic significance of metastatic lymph node ratio in node-positive colon carcinoma. Ann Surg Oncol 2007 May;14(5):1712–7. 23. Iqbal S, Stoehlmacher J, Lenz HJ. Tailored chemotherapy for colorectal cancer: a new approach to therapy. Cancer Invest 2004;22(5):762–73. 24. Westra JL, Plukker JT, Buys CH, et al. Genetic alterations in locally advanced stage II/III colon cancer: a search for prognostic markers. Clin Colorectal Cancer 2004;4(4):252–9. 25. Mader RM. Links between biology, prognosis and prediction of response to chemotherapy in colorectal cancer. Onkologie 2006;29(7): 334–41.
EJSO 34 (2008) 771e775
www.ejso.com
A study of lymph node ratio as a prognostic marker in colon cancer K. Derwinger*, G. Carlsson, B. Gustavsson ¨ stra, 41685 Gothenburg, Sweden Department of Surgery, Sahlgrenska University Hospital/O Accepted 1 November 2007 Available online 20 February 2008
Abstract Aim: The aim of this study was to evaluate and describe the lymph node ratio (LNR) as a prognostic parameter for patients with colon cancer. As lymphatic involvement is the key, focus was set at stage III disease. Interest was directed at the possibility of identifying high-risk groups and the clinical implementation and consequence. Method: The study was retrospective using a database of clinical data of all cancer patients treated at our unit. It has been continuous in registration, inclusion and update since 1999 including survival and clinical features. All patients (n ¼ 265) diagnosed with stage III colon cancer during 1999e2003 were included for the study. LNR was calculated and quartile groups were created. LNR and associated parameters were analysed towards 3-year disease-free survival (DFS). Basic patient data as well as surgery, pathology and postoperative treatment were taken into consideration. Results: Significant differences in disease-free survival were found for TNM N-status, tumour differentiation grade and LNR quartile group. There was a difference in 3-year DFS from 80% in LNR group 1 compared with less than 30% in group 4. These results were of prognostic interest both independently and in interaction with each other. High-risk groups could be identified and in the worst prognosis LNR group we also found a tendency towards more side effects with adjuvant chemotherapy. Conclusion: The lymph node ratio, the quota between the number of lymph node metastasis and assessed lymph nodes, is a highly significant ( p < 0.001) prognostic factor in stage III colon cancer. It can be an aid in identifying risk groups that could benefit from a more intense postoperative surveillance and possibly bring changes in adjuvant treatment strategy. More studies of clinical data, genetic and biochemical markers are needed in this patient group to understand the possible difference in tumour behaviour and tailor the treatment. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Colon cancer; Prognosis; Survival; Lymph node ratio
Introduction Colon cancer is one of the most common cancer forms in Sweden as well as in Western Europe and the incidence is slowly increasing.1 In being a common cancer form all information and knowledge that can assist in optimizing the process of staging, prognostics and choice of treatment strategy are of importance. The most important prognostic factor that influences long-term survival is the knowledge of the presence of lymph node involvement or distant metastases. Lymph node involvement is the main factor for including patients in adjuvant chemotherapeutic protocols.2 Failure to make an adequate staging assessment might deprive a patient of beneficial adjuvant therapy with resulting shorter survival. Simultaneously, the risk of over treating * Corresponding author. Tel.: þ46313434219; fax: þ46313435930. E-mail address: [email protected] (K. Derwinger). 0748-7983/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.ejso.2007.11.002
should also be considered with unnecessary risk of adverse effects for the patients. The most commonly used system for cancer staging is the TNM classification. It takes the main parameters of local growth, regional lymph node involvement and presence of distant spread into consideration.3 It is systematic and functional but within stage III disease there are prognostic limitations. One of these is that stage III is often treated as an entity and that the differences in prognosis within the group are not clear. This does affect the treatment and the follow-up and limits the possibility of a tailored strategy. The importance of the pathologist’s work and adherence to quality standards is also known.4,5 Another developing and interesting field that is rapidly expanding is that of biomolecular and genetic markers that might give information for prognosis and as aid in choosing the line of therapy.6 Many new techniques are also developed for metastasis detection.7 This will have an impact on cancer staging and
K. Derwinger et al. / EJSO 34 (2008) 771e775
772
primarily in differing between stage I/II and stage III and in the end on stage-specific survival. Node related factors, other than present ones, that can influence survival and thus can be of prognostic interest include lymph node size and distribution.8,9 Another possibility is the lymph node ratio (LNR) which is the quota between positive and assessed lymph node. This parameter is well described as a significant predictor in gastric cancer as well as in breast cancer and there is evidence of usefulness in colorectal cancer.10e12 The aim of this study was to evaluate and describe the lymph node ratio as a prognostic parameter for patients with colon cancer. As lymphatic involvement is the key, focus was set at stage III disease. Interest was directed at the possibility of identifying high-risk groups and the clinical implementation and consequence. The study is a retrospective analysis of data from a high volume surgical unit in a university hospital.
as this outcome parameter as it is an accepted way to reflect on the future long-term prognosis.14
Method
Results
Material and method
The patients and the surgery
¨ stra At the Department of Surgery, Sahlgrenska/O University Hospital, Gothenburg, we are continuously making a registration of detailed clinical and pathological data. The registration is consecutive since 1999 for all patients treated for colorectal cancer. Included in the database are also a continuous follow-up regarding treatment, recurrence and survival. In staging we use the TNM classification, version 5, 1997.13 During the period 1999e2003, 718 patients were diagnosed with colon cancer at our unit. For this study, we selected all patients (n ¼ 265) with stage III disease. The overall stage distribution during the period was stage (I) 9%, (II) 37%, (III) 36% and stage (IV) 18%. All the patients were treated following the same guidelines including postoperative surveillance. We studied basic clinical parameters such as sex, age, diagnosis, cancer location, operation as well as pathology data such as lymph nodes and differentiation grade. Interest was focused on number of nodes with metastasis growth correlated to the number of assessed nodes. Lymph node ratio was calculated as the quota between metastasis positive nodes and assessed lymph nodes. LNR groups were created on the basis of ratio quartiles. As prognostic instrument we used 3-year disease-free survival data
The average age in the material was 72 years. There was a slight female predominance. The preoperative staging at our unit was performed with chest X-ray and liver examination (CT or ultrasound) and was completed in 98% of elective patients. There was no difference in the spectra of surgical resections between the LNR groups with right hemicolectomy being the most common. Neither were there any differences in surgical panorama, age, sex or hospital stay when grouping either for LNR group, N-status or differentiation grade. The proportion of emergency procedures in the entire group was 24% with a variation from 20% in LNR group 1 to 29% in group 4. The median number of assessed nodes was 11 (range 4e160). The number of assessed nodes did not differ with any grouping but the number of metastatic nodes increased naturally with LNR group. Patient characteristics and pathology data details as presented in Table 1.
Statistical method We used the JMP 4/SAS software for statistical analysis. Basic patient demographic data were set by distribution statistics with ANOVA or contingency tables with Chi-square analysis. The KaplaneMeier method was used to calculate cumulative survival and Log rank test was used to compare survival differences between LNR groups. The same analyses were made for TNM N-status (node status: N1 or N2) as well as differentiation grade. We also performed a COX proportional hazard analysis to explore their significance and possible co-variation. All the analyses were also reexamined after exclusion of any patients in whom less than 12 nodes had been assessed, as by UICC guidelines, to ascertain that eventual results retained their significance.
Adjuvant therapy All stage III patients (n ¼ 265) were discussed in a multimodal treatment conference and if eligible were treated by
Table 1 Patients’ demographics and pathology data in LNR groups LNR group
LNR internal
1 2 3 4
0e0.125 0.126e0.266 0.267e0.450 0.451e1
Total
Number
63 71 66 65 265
Age, average (range)
Assessed nodes, medium (range)
Metastasis nodes, medium(range)
Differentiation grade (high/ medium/poor)
Adjuvant treatment (yes/no)
N-status (N1/N2)
70 72 71 71
13 12 10 11
1 2 3.5 6
5/45/13 3/48/20 1/34/31 0/31/34
39/24 44/27 36/30 38/27
63/0 53/18 32/34 8/57
8/159/98
157/108
156/109
(25e91) (44e90) (27e92) (32e90)
72 (25e92)
(8e160) (4e29) (5e31) (6e22)
11 (4e160)
(1e2) (1e5) (2e13) (3e20)
2 (1e20)
K. Derwinger et al. / EJSO 34 (2008) 771e775
our oncologists. There was no difference between the LNR groups in frequency of inclusion into adjuvant chemotherapy protocols (n ¼ 157). Neither was there any difference in the choice of regime. The most frequent protocol was with 5-FU and Leucovorin accounting for more than 95% of the treatments. There was a small tendency in LNR group 4 of having more side effects and thus more often than LNR groups 1e3 not completing the chemotherapeutic protocol. The difference was seen but does not reach significance ( p < 0.1). When examining the cohort that was not eligible for chemotherapy some data differed. As age is a limiting factor, the oldest patients are rarely eligible for adjuvant treatment. In this group (n ¼ 108) the mean age was 80 (range 41e92). There was no significant difference in the number of assessed nodes or tumour stages. Nor was there any significant difference in surgical panorama or hospital stay for the group. Age was the most frequent factor in this decision but concomitant disease also played a role. When separately examining the prognostic factors in this cohort LNR and N-status, although still significant, had less impact on cancer-specific survival whilst the differentiation grade was relatively more important ( p < 0.01). The nodes and differentiation The number of assessed positive nodes increased in LNR group to a significant level ( p < 0.05). There was no difference between any groupings in total number of assessed nodes. The analysis of survival and comparison between TNM system N1 (1e3 positive nodes, n ¼ 161) and N2 (more than three positive nodes, n ¼ 104) showed a significant difference in 3-year disease-free survival ( p < 0.001). There was also a co-variation between differentiation grade and node status both in prevalence and affecting survival. A worse differentiation grade was more common in N2-status than in N1. The aspect of tumour differentiation grade is also a significant ( p < 0.001) prognostic factor for survival where a worse (lower) differentiation grade corresponds to worse outcome. Between these groupings, there was no difference in analyses regarding sex, age or number of assessed nodes. When re-examining the same tests after exclusion of the cases where less than 12 nodes were assessed, we retained the same significance of our results. The differentiation grade also affected survival at a significant level ( p < 0.05) within each N group, giving an interesting co-variation.
773
Figure 1. Disease-free survival in LNR groups.
II levels. Though still in stage III, LNR group 4 in contrast had a far worse survival prognosis with a 29% 3-year DFS. A poor differentiation grade was more common in LNR group 4 whilst a high grade was uncommon in stage III cancer in our material. As LNR is a quota the N2 node status was more frequent in the higher LNR groups. In crossexamination of N-status (N1/N2) with differentiation we found significant ( p < 0.05) difference in survival in each category with the worst prognosis being N2 and poor differentiation grade. However, this did not fully match the worse prognosis of LNR group 4. In multivariate analysis LNR remained as a highly significant factor along with differentiation grade. Statistical analysis results are summarized in Table 2. Discussion Staging Staging is important in many aspects. It often decides the postoperative treatment and follow-up. Thus it affects the individual patient as the risk of harm from examination or drug side effect should be balanced by the risk of recurrence and chance of cure. It is also in some economic systems a needed aid in allocation of limited healthcare resources. Tools that can aid us in directing, individualizing and optimizing the postoperative process are of importance. The overall most significant prognostic indicator is the cancer stage either by Dukes or TNM stage IeIV.15 Absence of
LNR and survival The entire cohort (n ¼ 265) was divided into four groups after ratio quartiles. When comparing both overall survival and disease-free survival stratified by LNR group we found significant differences ( p < 0.001) (Fig. 1). In Cox proportional hazard analysis the risk ratio between LNR groups 1e2 was 1.8, LNR groups 2e3: 1.4 and groups 3e4: 2.0. LNR group 1 had a 3-year survival prognosis of 80% and thus well comparable to normally accepted stage
Table 2 Summary of statistical results Factor
Univariate, p-value
Multivariate (risk ratio (CI), p-value)
LNR N-status (N1/N2) Differentiation grade (medium/poor) Adjuvant therapy (yes/no)
<0.0001 <0.001 <0.001
10.6 (3.2e31.8), <0.0002 2.0 (1.1e3.9), <0.04 2.8 (1.2e6.1), <0.0006
<0.001
3.1 (2.1e4.7), <0.001
774
K. Derwinger et al. / EJSO 34 (2008) 771e775
lymph node involvement (N0 status) or distant spread (M0 status) means stage I or II disease and is thus strong positive prognostic markers. At the other end metastasis to other organs (M1-status), meaning stage IV disease, indicate a poor prognosis for long-term survival. In between lays stage III colon cancer which as shown is a heterogeneous group with a more variable outcome. Pathology A key factor in this process is the pathologist.16 It has been shown in several studies that their preparation, method and analysis will affect the report and thus also the clinical actions.17,18 The number of assessed lymph nodes needed for an adequate staging is set at 12 nodes in both national and international standard.19 This is a quality factor not to be neglected and a possible prognostic tool should be verified by data meeting this requirement as has been done here.20 Ratio calculation should be based on full node assessment but with focus on the nodes along the central spread vessels where a positive finding is more probable. This is often sub-specified in the reports from some pathology departments and might be of importance as the total number of assessed nodes is increasing. New possibilities of finding signs of cancer spread by PCR is also under development. It will be of interest though it will primarily increase the accuracy in defining between stage I/II and stage III disease. This in turn will affect the stage-specific survival prognostics through a possible stage migration. A ratio One possibility of further prognostic information within stage III disease is the lymph node ratio. It has a significant ( p < 0.001) impact in disease-free survival prognosis as shown in Fig. 1 and remained significant in a multivariate analysis. It can be an aid in finding risk groups where extra postoperative attention should be given. Possible factors contributing to the ratio prognostics are the number of assessed and positive nodes but also the differentiation grade as it covariates. With worse grade it is more frequent with more positive nodes and higher LNR group. The higher prevalence of emergency surgery in LNR group 4 can also contribute to worsening the outcome.21 However, separate analysis of the emergency cohort revealed the same ratio patterns and it could not by itself explain the prognostic difference. We believe that the ratio is a significant continuous variable in accordance with previous studies.18 Possibly it reflects tumour biology and the total cancer burden. Still, a grouping makes it more defined and facilitates the identification of risk groups. Other studies have tested quartile groupings with similar results and thus a possibility of comparison and confirmation.12,22 A higher number of groups make data more difficult to handle whilst fewer make risk categories difficult to distinguish.
Treatment and side effects The main line of chemotherapy for stage III disease is by 5-FU and Leucovorin, applying to more than 95% in this material.19 There was no difference in treatment between the ratio groups. Interesting factors that can alter or affect tumour malignancy potential and response to chemotherapy include genetics, bio-molecular markers and polymorphisms but these are under development and need further data and studies.23e25 Interesting is also the possible discussion of the treatment hazards and choice of therapy in a low-risk group. Whilst high-risk patients in group 4 might benefit from the most aggressive regimes, it is more difficult to advocate the same for low-risk patients in group 1. The development of tailored treatments might aid us in these questions. Regarding the cohort of patients that were not eligible for chemotherapy their overall prognosis was worse. This should be considered with the knowledge that age is a prognostic factor and that this group is older and more burdened by concomitant disease. Therefore, the clinical importance of the prognostic parameters and their significance is difficult to assess in this group. The study and beyond We believe that this material confirms the use of LNR as a possible prognostic tool. Our data and results are in line with and strengthen previous studies.12,22 In our opinion it does reflect the total tumour burden and biological properties. The study also confirms that the node related factors are and remain among the strongest prognostic parameters. The differentiation grade is of some importance and an interesting factor but in the literature its role is unclear. We were able to identify a high-risk patient group which could benefit from a more intense postoperative surveillance and possibly changes in the adjuvant treatment strategy. Additional studies must be made to find which drugs and regimes that can possibly give this group a better chance of long-term survival. One weakness in our material is the relatively small number of patients when compared to other studies. We believe this to be well compensated by the fact that the material is unselected and population based. We have good possibilities to survey and follow-up of the patients. The patients were treated at the same hospital mainly by the same staff and with same treatment strategies during this period. All were included, registered and treated using the same guidelines. Conclusion The lymph node ratio, the quota between the number of lymph node metastases and assessed lymph nodes, is a highly significant ( p < 0.001) prognostic factor in stage III colon cancer. It can be an aid in identifying risk groups that could benefit from a more intense postoperative surveillance and possibly changes in adjuvant treatment
K. Derwinger et al. / EJSO 34 (2008) 771e775
strategy. More studies of clinical data, genetic and biochemical markers are needed in this patient group to understand the possible difference in tumour behaviour and tailor the treatment. Conflict of interest The authors have no conflict of interest.
Acknowledgement We like to express our gratitude to the staff of our oncology laboratory unit assisting in the collection and registration of data and samples.
References 1. Oncological centre in Western Region of Sweden. Regional guidelines and statistics; 2005. 2. Glimelius B, Dahl O, Cedermark B, et al. Adjuvant chemotherapy in colorectal cancer: a joint analysis of randomised trials by the nordic gastrointestinal tumour adjuvant therapy group. Acta Oncol 2005; 44(8):904–12. 3. Kehoe J, Khatri VP. Staging and prognosis of colon cancer. Surg Oncol Clin N Am 2006;15(1):129–46. 4. Lemmens VE, Verheij CD, Janssen-Heijnen ML, et al. Mixed adherence to clinical practice guidelines for colorectal cancer in the southern Netherlands in 2002. Eur J Surg Oncol 2006;32(2):168–73. 5. Ratto C, Sofo L, Ippoliti M, et al. Accurate lymph-node detection in colorectal specimens resected for cancer is of prognostic significance. Dis Colon Rectum 1999;42(2):143–54. discussion 154e8. 6. Engwegen JY, Gast MC, Schellens JH, et al. Clinical proteomics: searching for better tumour markers with SELDI-TOF mass spectrometry. Trends Pharmacol Sci 2006;27(5):251–9. 7. Ohlsson L, Hammarstrom ML, Israelsson A, et al. Biomarker selection for detection of occult tumour cells in lymph nodes of colorectal cancer patients using real-time quantitative RT-PCR. Br J Cancer 2006; 95(2):218–25. 8. Kobayashi H, Ueno H, Hashiguchi Y, et al. Distribution of lymph node metastasis is a prognostic index in patients with stage III colon cancer. Surgery 2006;139(4):516–22.
775
9. Dhar DK, Yoshimura H, Kinukawa N, et al. Metastatic lymph node size and colorectal cancer prognosis. J Am Coll Surg 2005;200(1): 20–8. 10. Inoue K, Nakane Y, Iiyama H, et al. The superiority of ratio-based lymph node staging in gastric carcinoma. Ann Surg Oncol 2002; 9(1):27–34. 11. Voordeckers M, Vinh-Hung V, Van de Steene J, et al. The lymph node ratio as prognostic factor in node-positive breast cancer. Radiother Oncol 2004;70(3):225–30. 12. Berger AC, Sigurdson ER, LeVoyer T, et al. Colon cancer survival is associated with decreasing ratio of metastatic to examined lymph nodes. J Clin Oncol 2005;23(34):8706–12. 13. UICC/AJCC. Manual of TNM classification of malignant tumors. 5th ed.; 1997. 14. Sargent DJ, Wieand HS, Haller DG, et al. Disease-free survival versus overall survival as a primary end point for adjuvant colon cancer studies: individual patient data from 20,898 patients on 18 randomized trials. J Clin Oncol 2005;23(34):8664–70. 15. Wolters U, Stutzer H, Keller HW, et al. Colorectal cancer e a multivariate analysis of prognostic factors. Eur J Surg Oncol 1996;22(6):592–7. 16. Compton CC. Key issues in reporting common cancer specimens: problems in pathologic staging of colon cancer. Arch Pathol Lab Med 2006;130(3):318–24. 17. Lemmens VE, van Lijnschoten I, Janssen-Heijnen ML, et al. Pathology practice patterns affect lymph node evaluation and outcome of colon cancer: a population-based study. Ann Oncol 2006;17(12):1803–9. 18. Jestin P, Pahlman L, Glimelius B, et al. Cancer staging and survival in colon cancer is dependent on the quality of the pathologists’ specimen examination. Eur J Cancer 2005;41(14):2071–8. 19. National Board of Health in Sweden. National guidelines for colorectal cancer; 2006. 20. Derwinger K, Carlsson G, Gustavsson B. Stage migration in colorectal cancer related to improved lymph node assessment. Eur J Surg Oncol 2007 Sep;33(7):849–53. 21. Jestin P, Nilsson J, Heurgren M, et al. Emergency surgery for colonic cancer in a defined population. Br J Surg 2005;92(1):94–100. 22. Lee HY, Choi HJ, Park KJ, et al. Prognostic significance of metastatic lymph node ratio in node-positive colon carcinoma. Ann Surg Oncol 2007 May;14(5):1712–7. 23. Iqbal S, Stoehlmacher J, Lenz HJ. Tailored chemotherapy for colorectal cancer: a new approach to therapy. Cancer Invest 2004;22(5):762–73. 24. Westra JL, Plukker JT, Buys CH, et al. Genetic alterations in locally advanced stage II/III colon cancer: a search for prognostic markers. Clin Colorectal Cancer 2004;4(4):252–9. 25. Mader RM. Links between biology, prognosis and prediction of response to chemotherapy in colorectal cancer. Onkologie 2006;29(7): 334–41.