- Email: [email protected]
Lymph node size as a simple prognostic factor in node negative colon cancer and an alternative thesis to stage migration
The American Journal of Surgery (2015) -, -–-
Lymph node size as a simple prognostic factor in node negative colon cancer and an alternative thesis to stage migration Bruno Ma¨rkl, M.D.a,*,1, Tina Schaller, M.D.a,1, Yuriy Kokot, M.D.b, Katharina Endhardt, M.D.a, Hallie Kretsinger, M.D.a, Klaus Hirschbu¨hl, M.D.c, Georg Aumann, M.D.d, Gerhard Schenkirsch, M.D.e a
Institute of Pathology, Klinikum Augsburg, Augsburg, Germany; bIV Medical Clinic, Klinikum Augsburg, Augsburg, Germany; cII Medical Clinic, Klinikum Augsburg, Augsburg, Germany; d Department of Visceral and Transplantation Surgery, Klinikum Augsburg, Augsburg, Germany; and e Clinical and Population Based Cancer Registry Augsburg, Klinikum Augsburg, Augsburg, Germany
KEYWORDS: Colon cancer; Lymph node; Immune response; Size; Stage migration
Abstract BACKGROUND: Stage migration is an accepted explanation for the association between lymph node (LN) yield and outcome in colon cancer. To investigate whether the alternative thesis of immune response is more likely, we performed a retrospective study. METHODS: We enrolled 239 cases of node negative cancers, which were categorized according to the number of LNs with diameters larger than 5 mm (LN5) into the groups LN5-very low (0 to 1 LN5), LN5-low (2 to 5 LN5), and LN5-high (R6 LN5). RESULTS: Significant differences were found in pT3/4 cancers with median survival times of 40, 57, and 71 months (P 5 .022) in the LN5-very low, LN5-low, and LN5-high groups, respectively. Multivariable analysis revealed that LN5 number and infiltration type were independent prognostic factors. CONCLUSIONS: LN size is prognostic in node negative colon cancer. The correct explanation for outcome differences associated with LN harvest is probably the activation status of LNs. Ó 2015 Elsevier Inc. All rights reserved.
Lymph node (LN) staging is still of crucial importance for the prognosis estimation and therapy stratification in colon cancer. It has been confirmed by many studies that There were no relevant financial relationships or any sources of support in the form of grants, equipment, or drugs. The authors declare no conflicts of interest. * Corresponding author. Tel.: 149-821-4003199; fax: 149-821400173199. E-mail address: [email protected] Manuscript received March 1, 2015; revised manuscript May 8, 2015 1 Bruno Ma¨rkl and Tina Schaller contributed equally to this work. 0002-9610/$ - see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjsurg.2015.05.026
the number of evaluated LNs correlates with patient outcomes in nodal negative colon cancers.1–5 Missing LN metastases because of insufficient LN harvest could cause an incorrect stage classification, thereby confounding the survival analysis. A statistical stage migration effect, also known as Will Rogers phenomenon, is therefore the most frequently mentioned explanation for this association.6 Recently, we questioned this hypothesis. We hypothesized that the reported superior survival data actually are caused by an immunological effect. Activation of the immune system could result in an enlargement of LNs, which are
2
The American Journal of Surgery, Vol -, No -, - 2015
therefore easy to detect in high numbers. In the setting of conventional LN dissection, a high LN yield would not prevent understaging but indicates an activated state of the immune system. This could be a reasonable explanation of why advanced LN dissection techniques fail to improve the prognostic value of LN staging. To determine whether this is also true using the conventional LN staging technique, we performed a retrospective study enrolling a larger number of nodal negative cases with long follow-up.
sections were cut. All reactions were performed using a Ventana Benchmark Ultra system (Roche diagnostics, Mannheim, Germany). Prediluted mouse monoclonal antibodies including anti-MLH-1 (M1), anti-MSH2 (G2191129), anti-MSH6 (44), and the rabbit monoclonal antibody PMS2 (EPR3947) were used. The Ventana Vision detection system (Roche diagnostics) was used for the development of the reactions.
Patients and Methods Patients and follow-up Cases of nodal negative colon cancer were gathered from the files of the Institute of Pathology, Klinikum Augsburg between 2002 and 2005. Inclusion criteria were tumor-free resection margins, curative intention, and a follow-up of greater than or equal to 2 months. Follow-up data were provided by the Clinical and Population Based Cancer Registry Augsburg. The study was proven by the internal review board of the Klinikum Augsburg.
Histopathology During the time period mentioned above, LNs were dissected using the conventional manual dissection technique. LNs larger than 5 mm in diameter were bisected. After paraffin embedding at least 2 step sections were cut from each block. Then the slides were stained with hematoxylin and eosin (H&E). From each case, a representative slide of the tumor was re-evaluated by 2 independent investigators (B.M. and T.S.) blinded to the initial results. The evaluated features were pT-stage, histological type, invasion type according to Jass, tumor budding, lymphatic and venous invasion, lymphocytic infiltration, and Crohn-like inflammation. In discrepant cases, the slides were re-evaluated again, and a consensus decision was made. The invasion type was analyzed according to the description by Jass. Tumor budding was evaluated on H&E stained slides using the counting method. The cut-off value used was 30 buds/ 0.385 mm2. The lymphocytic infiltration was graded qualitatively using a 3-tailed system (no or low, moderate, and severe). No discrimination between intra- and peritumoral infiltration was made. Crohn-like inflammation again was evaluated according to Jass’ description. In mucinous, poorly differentiated and medullary carcinomas, additional immunohistochemical investigations were performed with an MLH1 antibody to determine acquired microsatellite instability (MSI). In cases where at least one of the Bethesda criteria was fulfilled, full MSI investigation including immunohistochemistry MLH1, MSH2, MSH6, PMS2, and molecular MSI testing were performed. For immunohistochemistry, 3 to 5 mm thin
Morphometry For the LN morphometry, a simple caliber (Fig. 1) was used on the basis of H&E stained slides. The investigator (Y.K.) counted the number of LNs with diameters greater than 5 mm in each case. Because of the significantly lower LN harvest because of the conventional dissection technique, the cut-off value was adapted. A harvest of greater than or equal to 6 LNs with diameters greater than 5 mm qualified a case as LN5-high (LN5h). All cases with 2 to 5 such LNs were regarded as LN5-low (LN5l). The remaining cases with zero or one LN greater than 5 mm were classified as LN5-very low (LN5vl). The study was approved by the internal review board of the Klinikum Augsburg.
Statistics Depending on the results of normality testing, metric values were compared using either the Student t test or the Mann–Whitney rank-sum test. Dichotomous data were analyzed with the chi-square test or Fisher’s exact test depending on the sample size. For the determination of the median follow-up time, the method of Schemper and Smith was used.7 Kaplan–Meier curves were created and log-rank regression analyses were performed to compare overall and tumor-related survival of the different groups. Multivariable analyses were performed using the Cox regression
Figure 1 Categorization of LNs was performed using a transparent caliber with a 5-mm hole based on H&E stained slides. This proved to be a feasible, very simple, and fast method.
B. Ma¨rkl et al.
Lymph node size in colon cancer
proportional hazards model. P values less than .05 were considered significant. All calculations were performed using the Sigma-Plot 11.0 software package (SYSTAT, Erkrath, Germany).
Results Lymph node measurement Measurement of the LNs using a caliber (Fig. 1) proved to be feasible. The number of LNs larger than 5 mm could be determined in all cases within less than 1 minute.
Case collection and clinicopathological characteristics
3 significantly more often on the right side of the colon than tumors belonging to the LN5l and LN5vl groups (87% vs 67% vs 37%, P , .001). Locally advanced tumors (pT3 or 4) were significantly more frequent in the LN5h group with 71% compared with 61% and 47% in the LN5l and LN5vl groups (P 5 .011), respectively. The other differences concern grading and mean age (Table 1). No association was found between the LN5 groups and occurrence of tumor infiltrating lymphocytes and/or Crohn-like inflammation (correlation coefficients .09 and .05, P values .148 and .380). The latter factors, however, showed marginally significant and significant associations with the LN count (correlation coefficients .125 and .171, P values .008 and .053).
Survival analysis A collection of 239 nodal negative colon cancer cases was enrolled. More than 5 LNs larger than 5 mm in diameter were found in 61 cases forming the LN5h group. The LN5l group consisted of 97 cases. The remaining 81 cases represented the LN5vl group. The clinicopathologic characteristics are summarized in Table 1. The mean number of LN greater than 5 mm in the entire collective was 4 6 4. There were considerable differences between the 3 groups. The mean LN count differed highly significantly between (Fig. 2) the groups with 11 6 4, 14 6 4, and 20 6 8 LNs in the LN5vl, LN5l, and LN5h groups, respectively (P , .001). Tumors with a large number of intermediate- to large-sized LNs (LN5h) were located highly
Table 1
The median follow-up time was 64 months (range 18 to 143 months). Survival analysis of the 3 LN5 groups independent from T-stage revealed a nonsignificant trend toward reduced median survival time of the LN5vl group compared with both the other groups (71 vs 76 months, P 5 .230) (Fig. 3A). However, after T-stage adjusting, significant differences were found in the group of locally advanced cases (pT3/4). The median survival times for the LN5vl, LN5l, and LN5h groups were 40, 57, and 71 months, respectively (P 5 .022) (Fig. 3B). A separate analysis restricted to cases with greater than or equal to 12 evaluated LN revealed mean (median not reached in
Clinicopathological data of the different size groups
Sex: male:female Mean age 6 SD Age , 50 Insufficient LN count (,12) Mean LN count 6 SD Conventional histology Mucinous differentiation Medullary differentiation Other differentiation Expansive type Infiltrative type High-grade tumor budding pT1/2 pT3/4 Low grade† High grade† Lymphatic invasion Venous invasion Distant metastases right location Left location
LN5vl (n 5 81) (%)
LN5l (n 5 97) (%)
LN5h (n 5 61) (%)
P value
1:.9 72 6 9 0 40 (50) 11 6 4 71 (88) 8 (10) 2 (2) 0 72 (90) 9 (10) 16 (20) 43 (53) 38 (47) 66 (81) 11 (14) 6 (7) 3 (4) 10 (12) 30 (37) 51 (63)
1:.64 71 6 10 3 (3) 20 (21) 14 6 4 86 (89) 7 (7) 2 (2) 2 (2) 85 (88) 12 (12) 24 (25) 38 (39) 59 (61) 63 (65) 29 (30) 2 (2) 3 (3) 10 (10) 67 (69) 30 (31)
1:.97 66 6 12 6 (10) 4 (7) 20 6 8 47 (77) 8 (13) 4 (7) 2 (3) 55 (90) 6 (10) 14 (23) 17 (28) 43 (71) 45 (74) 11 (18) 2 (3) 0 2 (3) 53 (87) 8 (13)
.353 .001 n.c. , .001 ,.001
.391* .885 .728 .011 .024 n.c. n.c. .145 , .001
LN 5 lymph node (h 5 high; l 5 low; vl 5 very low); n.c. 5 not calculated because of very small sample numbers; SD 5 standard deviation. *All differentiations other than conventional are grouped together. † Grading was not available in all cases (because of mucinous differentiation and unavailable paraffin blocks).
4
The American Journal of Surgery, Vol -, No -, - 2015 clinical course was found in cases with venous invasion (69 vs 36 months, P 5 .081). The other factors evaluateddhistological type and grading, tumor budding, tumor infiltrating lymphocytes, Crohn-like inflammation, and locationdwere not prognostic. After multivariable analysis, both LN5 grouping as well as invasion evaluation were found to be independently prognostic (hazard ratio 1.7 for both).
Comments
Figure 2 Mean LN counts in different 0LN5 categories. The error bars represent 1 standard deviation. There are highly significant differences between the groups clearly indicating an association between the LN size category and LN yield.
LN5vl group, n 5 105) survival times of 54, 68, and 83 months for the groups LN5vl, LN5l, and LN5h, respectively. These differences, however, were not significant (P 5 .627). Comparison of the expansive vs infiltrative invasion type according to Jass revealed a significantly worse outcome in the latter group (69 vs 43 month, P 5 .043). The very rare occurrence of lymphatic invasion was also associated with poor survival (69 vs 2 months, P , .001). A nonsignificant but clear trend toward an adverse
Despite huge efforts, the many aspects of LN staging are still debated controversially.8 The prognostic relevance of LN number, however, has been confirmed by many studies including a systematic review.1,3–5 A stage migration effectdalso known as Will Rogers phenomenondin which understaging is attributed to missing positive LNs, has been postulated as the most likely explanation.6 Our group could confirm that understaging can occur even in sufficiently staged cases.9 However, it seems questionable whether this effect can really explain the adverse outcome reported by Swanson et al4 in which stage II colon cancers with less than 5 examined that LNs had a 5-year survival rate of only 69%. Based on their data, we calculated that about 90% of the stage II cases must have had occult LN metastases to explain the outcome difference compared with cases with high LN yields. We addressed these issues in 2 previously published studies. In both studies, the methylene blue–assisted LN dissecting technique was used. On one hand,
Figure 3 Survival analysis comparing the overall survival of the 3 groups LN5vl (very low), LN5l (low), and LN5h (high). (A) Kaplan– Meier curve including all T-stages. There is an insignificant trend toward more aggressive behavior of cases with fewer LNs of intermediate to large size. (B) The restriction to locally advanced (pT3/4) cases reveals significant survival differences with poor outcome of the LN5vl group.
B. Ma¨rkl et al.
Lymph node size in colon cancer
we showed that excellent LN yield is not associated with a higher rate of LN metastases.10 On the other hand, a high number of intermediate- and large-sized LNs is associated with an increased mean number of detected LNs and a favorable outcome.11 Re-evaluation of this collection revealed an adverse outcome in those patients with a paucity of LNs larger than 5 mm (work in progress). This indicates that LN size as an indicator of immune system function is the true prognostic feature hidden behind the LN number. In this study, we addressed the topic of LN size in a conventionally staged group of nodal negative colon cancer cases. In comparison with our former study,11 which enrolled cases with high LN yield, we had to adjust the cut-off for the LN5h group from 7 to 6 LNs. The cut-off for the very low group with zero or one LN was identical to previous investigation. We assumed that the latter group has a poor outcome. In the analysis of our data, we saw a highly significant association between the 3 size groups, mean LN counts (Fig. 2) and the rate of insufficient LN staging (Table 1) with especially poor yields in the LN5vl group. While overall survival did not differ significantly between the 3 size groups, in analyzing the whole study collection (Fig. 3A), significant differences were found while evaluating locally advanced pT3/4 cases (Fig. 3B). Although the prognostic effect is somewhat smaller, this generally confirms the findings of the former study. Methylene blue–assisted LN dissection ensures perfect staging, which is perhaps more accurate in differentiating the LN size groups. As expected, the LN5vl group had a relatively poor median survival of only 40 months and a 5-year survival rate of 40% compared with 71 months and 64% in the LN5h group. It is remarkable that only the invasion type provided additional and independent prognostic information in a relevant number of cases. Lymphatic permeation was also associated with an adverse clinical course. However, because of its very rare occurrence in our collection (Table 1), we did not include it in the multivariable analysis. In our opinion, our data support our hypothesis that an immunological effect instead of stage migration is the true reason for the prognostic effect of LN count in colon cancer. It very probably represents the confounder postulated by Simunovic and Baxter12 in an editorial commenting on the results of Wong et al13 who found no survival differences between patients treated in hospitals with low vs high numbers of examined LNs. On the other hand, when analyzing all cases together regardless of the corresponding hospitals, LN number was prognostic again. Because LN enlargement enhances the detectability of LNs, such casesdregardless of the diligence of the pathologistdwould be associated with a high LN number and probably with a favorable outcome. An association between LN size and retrieval was also reported recently by Sloothaak et al.14 George et al15 also doubted the socalled Will Rogers phenomenon after analyzing a large cohort of more than 3,000 patients and assumed that immune response with LN enlargement is responsible for the better outcome in cases with many investigated LNs.
5 They found an association between tumor infiltrating lymphocytes and LN number. In concordance with that, we found that Crohn-like inflammation (P 5 .008) and tumor infiltrating LNs (P 5 .053) were associated with the LN count. However, both factors themselves were not prognostic. The results of several other studies failed to provide evidence that stage migration is the relevant mechanism for the prognostic effect of LN count. Recently, O’Shea et al16 reported an increase in investigated LNs over time in a large cohort but no increase in nodal positive cases. The same has been found by others before.17–21 In contrast, Sjo et al found an increase in stage III cases with improved LN retrieval. However, it has to be stated that the initial rate of 25% was very low.22 Very recently, Bla¨ker et al found no survival difference between cases with low (,12) and high LN counts (R12) in a collection of 1,899 patients with a 78% rate of sufficiently staged cases. The authors concluded that the particularly diligent LN dissection in their collection avoided false LN negativity. Under such circumstances, cases with less than 12 investigated LNs would not be prone to understaging. Reaching a stage III/ II ratio of 1, which is the highest reported in the literature, supports the stage migration thesis.23 As mentioned before, our own previously published data do not support this. We found no significant difference in LN positivity rates when comparing 2 collections with sufficiency rates of 98% and 62%.10 Pathologists seem to be very effective in detecting positive LNs even when only a few LNs are evaluated. In an experimental design, we found that in 85% of cases, a first LN metastasis was detected by evaluating only 9 LNs.10 Chen et al24 found in their study that re-evaluation of 83 primary insufficiently staged colorectal cancer cases induced upstaging in only one single case from N1 to N2. In our opinion, all these findings support our thesis of a correlation among LN yield, LN size, and immune response. The latter is probably the true biological factor that influences the clinical course. PT3/4 cases with no or only one LN larger than 5 mm showed a particularly adverse clinical course with a median survival time of only 40 months. Pihl et al already investigated the prognostic role of LN hyperplasia and size in colorectal cancer in 1980. They found significant associations between these factors and outcome.25 It is well known that about 20% of stage II colon cancers have an adverse outcome. However, the benefit of adjuvant chemotherapy is debated. In general, it is restricted to highrisk situations including obstruction, perforation, inadequate LN sampling, or T4 disease.26–29 Both histopathological and molecular markers have been identified and suggested30,31 to stratify colon cancer. However, none of these additional factors achieved recommendation in national and international guidelines.32,33 Our data show that LN size analysis offers a new way to identify cases with a high risk for aggressive behavior. In this context, it seems somewhat ironical that the quality-related recommendations force pathologists to investigate 12 LNs under all circumstances.
6
The American Journal of Surgery, Vol -, No -, - 2015
Because this can be achieved by extensively embedding mesenterial fat and counting minute LNs, cases with ineffective immune response are masked. We strongly believe that a standardized LN dissection technique like methylene blue–assisted dissection or fat clearance combined with the almost ridiculously simple LN size analysis could help to overcome these problems.
14. Sloothaak DA, Grewal S, Doornewaard H, et al. Lymph node size as a predictor of lymphatic staging in colonic cancer. Br J Surg 2014;101: 701–6. 15. George S, Primrose J, Talbot R, et al. Will Rogers revisited: prospective observational study of survival of 3592 patients with colorectal cancer according to number of nodes examined by pathologists. Br J Cancer 2006;95:841–7. 16. O’Shea A, Aly O, Parnaby CN, et al. Increased lymph node yield in colorectal cancer is not necessarily associated with a greater number of lymph node positive cancers. PLoS One 2014;9:e104991. 17. Parsons HM, Tuttle TM, Kuntz KM, et al. Association between lymph node evaluation for colon cancer and node positivity over the past 20 years. JAMA 2011;306:1089–97. 18. Porter GA, Urquhart R, Bu J, et al. Improving nodal harvest in colorectal cancer: so what? Ann Surg Oncol 2012;19:1066–73. 19. Storli K, Sondenaa K, Furnes B, et al. Improved lymph node harvest from resected colon cancer specimens did not cause upstaging from TNM stage II to III. World J Surg 2011;35:2796–803. 20. Moore J, Hyman N, Callas P, et al. Staging error does not explain the relationship between the number of lymph nodes in a colon cancer specimen and survival. Surgery 2010;147:358–65. 21. van Erning FN, Crolla RM, Rutten HJ, et al. No change in lymph node positivity rate despite increased lymph node yield and improved survival in colon cancer. Eur J Cancer 2014;50:3221–9. 22. Sjo OH, Merok MA, Svindland A, et al. Prognostic impact of lymph node harvest and lymph node ratio in patients with colon cancer. Dis Colon Rectum 2012;55:307–15. 23. Bla¨ker H, Hildebrandt B, Riess H, et al. Lymph node count and prognosis in colorectal cancer: the influence of examination quality. Int J Cancer 2015;136:1957–66. 24. Chen L, Kalady MF, Goldblum J, et al. Does reevaluation of colorectal cancers with inadequate nodal yield lead to stage migration or the identification of metastatic lymph nodes? Dis Colon Rectum 2014; 57:432–7. 25. Pihl E, Nairn RC, Milne BJ, et al. Lymphoid hyperplasia: a major prognostic feature in 519 cases of colorectal carcinoma. Am J Pathol 1980;100:469–80. 26. Figueredo A, Coombes ME, Mukherjee S. Adjuvant therapy for completely resected stage II colon cancer. Cochrane Database Syst Rev; 2008:CD005390. 27. Lin HH, Chang YY, Lin JK, et al. The role of adjuvant chemotherapy in stage II colorectal cancer patients. Int J Colorectal Dis 2014;29: 1237–43. 28. Gill S, Loprinzi CL, Sargent DJ, et al. Pooled analysis of fluorouracilbased adjuvant therapy for stage II and III colon cancer: who benefits and by how much? J Clin Oncol 2004;22:1797–806. 29. Johnston PG. Stage II colorectal cancer: to treat or not to treat. Oncologist 2005;10:332–4. 30. Schneider NI, Langner C. Prognostic stratification of colorectal cancer patients: current perspectives. Cancer Manag Res 2014;6:291–300. 31. Zlobec I, Lugli A. Prognostic and predictive factors in colorectal cancer. J Clin Pathol 2008;61:561–9. 32. Pox C, Aretz S, Bischoff SC, et al. S3-Leitlinie Kolorektales Karzinom, Langversion 1.0. Available at: http://leitlinienprogrammonkologie.de/leitlinien.7.0.html. Accessed November 23, 2014. 33. van de Velde CJ, Boelens PG, Borras JM, et al. EURECCA colorectal: multidisciplinary management: European consensus conference colon & rectum. Eur J Cancer 2014;50:1.e1–1.e34.
Acknowledgment The authors are thankful to Kathrin Ferstl-Blahetek and Elfriede Schwarz for excellent technical assistance.
References 1. Chang GJ, Rodriguez-Bigas MA, Skibber JM, et al. Lymph node evaluation and survival after curative resection of colon cancer: systematic review. J Natl Cancer Inst 2007;99:433–41. 2. Edler D, Ohrling K, Hallstrom M, et al. The number of analyzed lymph nodesda prognostic factor in colorectal cancer. Acta Oncol 2007;46:975–81. 3. Goldstein NS. Lymph node recoveries from 2427 pT3 colorectal resection specimens spanning 45 years: recommendations for a minimum number of recovered lymph nodes based on predictive probabilities. Am J Surg Pathol 2002;26:179–89. 4. Swanson RS, Compton CC, Stewart AK, et al. The prognosis of T3N0 colon cancer is dependent on the number of lymph nodes examined. Ann Surg Oncol 2003;10:65–71. 5. Lykke J, Roikjaer O, Jess P, Danish Colorectal Cancer Group. The relation between lymph node status and survival in Stage I-III colon cancer: results from a prospective nationwide cohort study. Colorectal Dis 2013;15:559–65. 6. Feinstein AR, Sosin DM, Wells CK. The Will Rogers phenomenond stage migration and new diagnostic techniques as a source of misleading statistics for survival in cancer. N Engl J Med 1985;312:1604–8. 7. Schemper M, Smith TL. A note on quantifying follow-up in studies of failure time. Control Clin Trials 1996;17:343–6. 8. Resch A, Langner C. Lymph node staging in colorectal cancer: old controversies and recent advances. World J Gastroenterol 2013;19:8515–26. 9. Ma¨rkl B, Kerwel TG, Ja¨hnig HG, et al. Methylene blue-assisted lymph node dissection in colon specimens: a prospective, randomized study. Am J Clin Pathol 2008;130:913–9. 10. Ma¨rkl B, Schaller T, Krammer I, et al. Methylene blue-assisted lymph node dissection technique is not associated with an increased detection of lymph node metastases in colorectal cancer. Mod Pathol 2013;26:1246–54. 11. Ma¨rkl B, Ro¨ssle J, Arnholdt HM, et al. The clinical significance of lymph node size in colon cancer. Mod Pathol 2012;25:1413–22. 12. Simunovic M, Baxter NN. Lymph node counts in colon cancer surgery: lessons for users of quality indicators. JAMA 2007;298: 2194–5. 13. Wong SL, Ji H, Hollenbeck BK, et al. Hospital lymph node examination rates and survival after resection for colon cancer. JAMA 2007; 298:2149–54.
Lymph node size as a simple prognostic factor in node negative colon cancer and an alternative thesis to stage migration Bruno Ma¨rkl, M.D.a,*,1, Tina Schaller, M.D.a,1, Yuriy Kokot, M.D.b, Katharina Endhardt, M.D.a, Hallie Kretsinger, M.D.a, Klaus Hirschbu¨hl, M.D.c, Georg Aumann, M.D.d, Gerhard Schenkirsch, M.D.e a
Institute of Pathology, Klinikum Augsburg, Augsburg, Germany; bIV Medical Clinic, Klinikum Augsburg, Augsburg, Germany; cII Medical Clinic, Klinikum Augsburg, Augsburg, Germany; d Department of Visceral and Transplantation Surgery, Klinikum Augsburg, Augsburg, Germany; and e Clinical and Population Based Cancer Registry Augsburg, Klinikum Augsburg, Augsburg, Germany
KEYWORDS: Colon cancer; Lymph node; Immune response; Size; Stage migration
Abstract BACKGROUND: Stage migration is an accepted explanation for the association between lymph node (LN) yield and outcome in colon cancer. To investigate whether the alternative thesis of immune response is more likely, we performed a retrospective study. METHODS: We enrolled 239 cases of node negative cancers, which were categorized according to the number of LNs with diameters larger than 5 mm (LN5) into the groups LN5-very low (0 to 1 LN5), LN5-low (2 to 5 LN5), and LN5-high (R6 LN5). RESULTS: Significant differences were found in pT3/4 cancers with median survival times of 40, 57, and 71 months (P 5 .022) in the LN5-very low, LN5-low, and LN5-high groups, respectively. Multivariable analysis revealed that LN5 number and infiltration type were independent prognostic factors. CONCLUSIONS: LN size is prognostic in node negative colon cancer. The correct explanation for outcome differences associated with LN harvest is probably the activation status of LNs. Ó 2015 Elsevier Inc. All rights reserved.
Lymph node (LN) staging is still of crucial importance for the prognosis estimation and therapy stratification in colon cancer. It has been confirmed by many studies that There were no relevant financial relationships or any sources of support in the form of grants, equipment, or drugs. The authors declare no conflicts of interest. * Corresponding author. Tel.: 149-821-4003199; fax: 149-821400173199. E-mail address: [email protected] Manuscript received March 1, 2015; revised manuscript May 8, 2015 1 Bruno Ma¨rkl and Tina Schaller contributed equally to this work. 0002-9610/$ - see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjsurg.2015.05.026
the number of evaluated LNs correlates with patient outcomes in nodal negative colon cancers.1–5 Missing LN metastases because of insufficient LN harvest could cause an incorrect stage classification, thereby confounding the survival analysis. A statistical stage migration effect, also known as Will Rogers phenomenon, is therefore the most frequently mentioned explanation for this association.6 Recently, we questioned this hypothesis. We hypothesized that the reported superior survival data actually are caused by an immunological effect. Activation of the immune system could result in an enlargement of LNs, which are
2
The American Journal of Surgery, Vol -, No -, - 2015
therefore easy to detect in high numbers. In the setting of conventional LN dissection, a high LN yield would not prevent understaging but indicates an activated state of the immune system. This could be a reasonable explanation of why advanced LN dissection techniques fail to improve the prognostic value of LN staging. To determine whether this is also true using the conventional LN staging technique, we performed a retrospective study enrolling a larger number of nodal negative cases with long follow-up.
sections were cut. All reactions were performed using a Ventana Benchmark Ultra system (Roche diagnostics, Mannheim, Germany). Prediluted mouse monoclonal antibodies including anti-MLH-1 (M1), anti-MSH2 (G2191129), anti-MSH6 (44), and the rabbit monoclonal antibody PMS2 (EPR3947) were used. The Ventana Vision detection system (Roche diagnostics) was used for the development of the reactions.
Patients and Methods Patients and follow-up Cases of nodal negative colon cancer were gathered from the files of the Institute of Pathology, Klinikum Augsburg between 2002 and 2005. Inclusion criteria were tumor-free resection margins, curative intention, and a follow-up of greater than or equal to 2 months. Follow-up data were provided by the Clinical and Population Based Cancer Registry Augsburg. The study was proven by the internal review board of the Klinikum Augsburg.
Histopathology During the time period mentioned above, LNs were dissected using the conventional manual dissection technique. LNs larger than 5 mm in diameter were bisected. After paraffin embedding at least 2 step sections were cut from each block. Then the slides were stained with hematoxylin and eosin (H&E). From each case, a representative slide of the tumor was re-evaluated by 2 independent investigators (B.M. and T.S.) blinded to the initial results. The evaluated features were pT-stage, histological type, invasion type according to Jass, tumor budding, lymphatic and venous invasion, lymphocytic infiltration, and Crohn-like inflammation. In discrepant cases, the slides were re-evaluated again, and a consensus decision was made. The invasion type was analyzed according to the description by Jass. Tumor budding was evaluated on H&E stained slides using the counting method. The cut-off value used was 30 buds/ 0.385 mm2. The lymphocytic infiltration was graded qualitatively using a 3-tailed system (no or low, moderate, and severe). No discrimination between intra- and peritumoral infiltration was made. Crohn-like inflammation again was evaluated according to Jass’ description. In mucinous, poorly differentiated and medullary carcinomas, additional immunohistochemical investigations were performed with an MLH1 antibody to determine acquired microsatellite instability (MSI). In cases where at least one of the Bethesda criteria was fulfilled, full MSI investigation including immunohistochemistry MLH1, MSH2, MSH6, PMS2, and molecular MSI testing were performed. For immunohistochemistry, 3 to 5 mm thin
Morphometry For the LN morphometry, a simple caliber (Fig. 1) was used on the basis of H&E stained slides. The investigator (Y.K.) counted the number of LNs with diameters greater than 5 mm in each case. Because of the significantly lower LN harvest because of the conventional dissection technique, the cut-off value was adapted. A harvest of greater than or equal to 6 LNs with diameters greater than 5 mm qualified a case as LN5-high (LN5h). All cases with 2 to 5 such LNs were regarded as LN5-low (LN5l). The remaining cases with zero or one LN greater than 5 mm were classified as LN5-very low (LN5vl). The study was approved by the internal review board of the Klinikum Augsburg.
Statistics Depending on the results of normality testing, metric values were compared using either the Student t test or the Mann–Whitney rank-sum test. Dichotomous data were analyzed with the chi-square test or Fisher’s exact test depending on the sample size. For the determination of the median follow-up time, the method of Schemper and Smith was used.7 Kaplan–Meier curves were created and log-rank regression analyses were performed to compare overall and tumor-related survival of the different groups. Multivariable analyses were performed using the Cox regression
Figure 1 Categorization of LNs was performed using a transparent caliber with a 5-mm hole based on H&E stained slides. This proved to be a feasible, very simple, and fast method.
B. Ma¨rkl et al.
Lymph node size in colon cancer
proportional hazards model. P values less than .05 were considered significant. All calculations were performed using the Sigma-Plot 11.0 software package (SYSTAT, Erkrath, Germany).
Results Lymph node measurement Measurement of the LNs using a caliber (Fig. 1) proved to be feasible. The number of LNs larger than 5 mm could be determined in all cases within less than 1 minute.
Case collection and clinicopathological characteristics
3 significantly more often on the right side of the colon than tumors belonging to the LN5l and LN5vl groups (87% vs 67% vs 37%, P , .001). Locally advanced tumors (pT3 or 4) were significantly more frequent in the LN5h group with 71% compared with 61% and 47% in the LN5l and LN5vl groups (P 5 .011), respectively. The other differences concern grading and mean age (Table 1). No association was found between the LN5 groups and occurrence of tumor infiltrating lymphocytes and/or Crohn-like inflammation (correlation coefficients .09 and .05, P values .148 and .380). The latter factors, however, showed marginally significant and significant associations with the LN count (correlation coefficients .125 and .171, P values .008 and .053).
Survival analysis A collection of 239 nodal negative colon cancer cases was enrolled. More than 5 LNs larger than 5 mm in diameter were found in 61 cases forming the LN5h group. The LN5l group consisted of 97 cases. The remaining 81 cases represented the LN5vl group. The clinicopathologic characteristics are summarized in Table 1. The mean number of LN greater than 5 mm in the entire collective was 4 6 4. There were considerable differences between the 3 groups. The mean LN count differed highly significantly between (Fig. 2) the groups with 11 6 4, 14 6 4, and 20 6 8 LNs in the LN5vl, LN5l, and LN5h groups, respectively (P , .001). Tumors with a large number of intermediate- to large-sized LNs (LN5h) were located highly
Table 1
The median follow-up time was 64 months (range 18 to 143 months). Survival analysis of the 3 LN5 groups independent from T-stage revealed a nonsignificant trend toward reduced median survival time of the LN5vl group compared with both the other groups (71 vs 76 months, P 5 .230) (Fig. 3A). However, after T-stage adjusting, significant differences were found in the group of locally advanced cases (pT3/4). The median survival times for the LN5vl, LN5l, and LN5h groups were 40, 57, and 71 months, respectively (P 5 .022) (Fig. 3B). A separate analysis restricted to cases with greater than or equal to 12 evaluated LN revealed mean (median not reached in
Clinicopathological data of the different size groups
Sex: male:female Mean age 6 SD Age , 50 Insufficient LN count (,12) Mean LN count 6 SD Conventional histology Mucinous differentiation Medullary differentiation Other differentiation Expansive type Infiltrative type High-grade tumor budding pT1/2 pT3/4 Low grade† High grade† Lymphatic invasion Venous invasion Distant metastases right location Left location
LN5vl (n 5 81) (%)
LN5l (n 5 97) (%)
LN5h (n 5 61) (%)
P value
1:.9 72 6 9 0 40 (50) 11 6 4 71 (88) 8 (10) 2 (2) 0 72 (90) 9 (10) 16 (20) 43 (53) 38 (47) 66 (81) 11 (14) 6 (7) 3 (4) 10 (12) 30 (37) 51 (63)
1:.64 71 6 10 3 (3) 20 (21) 14 6 4 86 (89) 7 (7) 2 (2) 2 (2) 85 (88) 12 (12) 24 (25) 38 (39) 59 (61) 63 (65) 29 (30) 2 (2) 3 (3) 10 (10) 67 (69) 30 (31)
1:.97 66 6 12 6 (10) 4 (7) 20 6 8 47 (77) 8 (13) 4 (7) 2 (3) 55 (90) 6 (10) 14 (23) 17 (28) 43 (71) 45 (74) 11 (18) 2 (3) 0 2 (3) 53 (87) 8 (13)
.353 .001 n.c. , .001 ,.001
.391* .885 .728 .011 .024 n.c. n.c. .145 , .001
LN 5 lymph node (h 5 high; l 5 low; vl 5 very low); n.c. 5 not calculated because of very small sample numbers; SD 5 standard deviation. *All differentiations other than conventional are grouped together. † Grading was not available in all cases (because of mucinous differentiation and unavailable paraffin blocks).
4
The American Journal of Surgery, Vol -, No -, - 2015 clinical course was found in cases with venous invasion (69 vs 36 months, P 5 .081). The other factors evaluateddhistological type and grading, tumor budding, tumor infiltrating lymphocytes, Crohn-like inflammation, and locationdwere not prognostic. After multivariable analysis, both LN5 grouping as well as invasion evaluation were found to be independently prognostic (hazard ratio 1.7 for both).
Comments
Figure 2 Mean LN counts in different 0LN5 categories. The error bars represent 1 standard deviation. There are highly significant differences between the groups clearly indicating an association between the LN size category and LN yield.
LN5vl group, n 5 105) survival times of 54, 68, and 83 months for the groups LN5vl, LN5l, and LN5h, respectively. These differences, however, were not significant (P 5 .627). Comparison of the expansive vs infiltrative invasion type according to Jass revealed a significantly worse outcome in the latter group (69 vs 43 month, P 5 .043). The very rare occurrence of lymphatic invasion was also associated with poor survival (69 vs 2 months, P , .001). A nonsignificant but clear trend toward an adverse
Despite huge efforts, the many aspects of LN staging are still debated controversially.8 The prognostic relevance of LN number, however, has been confirmed by many studies including a systematic review.1,3–5 A stage migration effectdalso known as Will Rogers phenomenondin which understaging is attributed to missing positive LNs, has been postulated as the most likely explanation.6 Our group could confirm that understaging can occur even in sufficiently staged cases.9 However, it seems questionable whether this effect can really explain the adverse outcome reported by Swanson et al4 in which stage II colon cancers with less than 5 examined that LNs had a 5-year survival rate of only 69%. Based on their data, we calculated that about 90% of the stage II cases must have had occult LN metastases to explain the outcome difference compared with cases with high LN yields. We addressed these issues in 2 previously published studies. In both studies, the methylene blue–assisted LN dissecting technique was used. On one hand,
Figure 3 Survival analysis comparing the overall survival of the 3 groups LN5vl (very low), LN5l (low), and LN5h (high). (A) Kaplan– Meier curve including all T-stages. There is an insignificant trend toward more aggressive behavior of cases with fewer LNs of intermediate to large size. (B) The restriction to locally advanced (pT3/4) cases reveals significant survival differences with poor outcome of the LN5vl group.
B. Ma¨rkl et al.
Lymph node size in colon cancer
we showed that excellent LN yield is not associated with a higher rate of LN metastases.10 On the other hand, a high number of intermediate- and large-sized LNs is associated with an increased mean number of detected LNs and a favorable outcome.11 Re-evaluation of this collection revealed an adverse outcome in those patients with a paucity of LNs larger than 5 mm (work in progress). This indicates that LN size as an indicator of immune system function is the true prognostic feature hidden behind the LN number. In this study, we addressed the topic of LN size in a conventionally staged group of nodal negative colon cancer cases. In comparison with our former study,11 which enrolled cases with high LN yield, we had to adjust the cut-off for the LN5h group from 7 to 6 LNs. The cut-off for the very low group with zero or one LN was identical to previous investigation. We assumed that the latter group has a poor outcome. In the analysis of our data, we saw a highly significant association between the 3 size groups, mean LN counts (Fig. 2) and the rate of insufficient LN staging (Table 1) with especially poor yields in the LN5vl group. While overall survival did not differ significantly between the 3 size groups, in analyzing the whole study collection (Fig. 3A), significant differences were found while evaluating locally advanced pT3/4 cases (Fig. 3B). Although the prognostic effect is somewhat smaller, this generally confirms the findings of the former study. Methylene blue–assisted LN dissection ensures perfect staging, which is perhaps more accurate in differentiating the LN size groups. As expected, the LN5vl group had a relatively poor median survival of only 40 months and a 5-year survival rate of 40% compared with 71 months and 64% in the LN5h group. It is remarkable that only the invasion type provided additional and independent prognostic information in a relevant number of cases. Lymphatic permeation was also associated with an adverse clinical course. However, because of its very rare occurrence in our collection (Table 1), we did not include it in the multivariable analysis. In our opinion, our data support our hypothesis that an immunological effect instead of stage migration is the true reason for the prognostic effect of LN count in colon cancer. It very probably represents the confounder postulated by Simunovic and Baxter12 in an editorial commenting on the results of Wong et al13 who found no survival differences between patients treated in hospitals with low vs high numbers of examined LNs. On the other hand, when analyzing all cases together regardless of the corresponding hospitals, LN number was prognostic again. Because LN enlargement enhances the detectability of LNs, such casesdregardless of the diligence of the pathologistdwould be associated with a high LN number and probably with a favorable outcome. An association between LN size and retrieval was also reported recently by Sloothaak et al.14 George et al15 also doubted the socalled Will Rogers phenomenon after analyzing a large cohort of more than 3,000 patients and assumed that immune response with LN enlargement is responsible for the better outcome in cases with many investigated LNs.
5 They found an association between tumor infiltrating lymphocytes and LN number. In concordance with that, we found that Crohn-like inflammation (P 5 .008) and tumor infiltrating LNs (P 5 .053) were associated with the LN count. However, both factors themselves were not prognostic. The results of several other studies failed to provide evidence that stage migration is the relevant mechanism for the prognostic effect of LN count. Recently, O’Shea et al16 reported an increase in investigated LNs over time in a large cohort but no increase in nodal positive cases. The same has been found by others before.17–21 In contrast, Sjo et al found an increase in stage III cases with improved LN retrieval. However, it has to be stated that the initial rate of 25% was very low.22 Very recently, Bla¨ker et al found no survival difference between cases with low (,12) and high LN counts (R12) in a collection of 1,899 patients with a 78% rate of sufficiently staged cases. The authors concluded that the particularly diligent LN dissection in their collection avoided false LN negativity. Under such circumstances, cases with less than 12 investigated LNs would not be prone to understaging. Reaching a stage III/ II ratio of 1, which is the highest reported in the literature, supports the stage migration thesis.23 As mentioned before, our own previously published data do not support this. We found no significant difference in LN positivity rates when comparing 2 collections with sufficiency rates of 98% and 62%.10 Pathologists seem to be very effective in detecting positive LNs even when only a few LNs are evaluated. In an experimental design, we found that in 85% of cases, a first LN metastasis was detected by evaluating only 9 LNs.10 Chen et al24 found in their study that re-evaluation of 83 primary insufficiently staged colorectal cancer cases induced upstaging in only one single case from N1 to N2. In our opinion, all these findings support our thesis of a correlation among LN yield, LN size, and immune response. The latter is probably the true biological factor that influences the clinical course. PT3/4 cases with no or only one LN larger than 5 mm showed a particularly adverse clinical course with a median survival time of only 40 months. Pihl et al already investigated the prognostic role of LN hyperplasia and size in colorectal cancer in 1980. They found significant associations between these factors and outcome.25 It is well known that about 20% of stage II colon cancers have an adverse outcome. However, the benefit of adjuvant chemotherapy is debated. In general, it is restricted to highrisk situations including obstruction, perforation, inadequate LN sampling, or T4 disease.26–29 Both histopathological and molecular markers have been identified and suggested30,31 to stratify colon cancer. However, none of these additional factors achieved recommendation in national and international guidelines.32,33 Our data show that LN size analysis offers a new way to identify cases with a high risk for aggressive behavior. In this context, it seems somewhat ironical that the quality-related recommendations force pathologists to investigate 12 LNs under all circumstances.
6
The American Journal of Surgery, Vol -, No -, - 2015
Because this can be achieved by extensively embedding mesenterial fat and counting minute LNs, cases with ineffective immune response are masked. We strongly believe that a standardized LN dissection technique like methylene blue–assisted dissection or fat clearance combined with the almost ridiculously simple LN size analysis could help to overcome these problems.
14. Sloothaak DA, Grewal S, Doornewaard H, et al. Lymph node size as a predictor of lymphatic staging in colonic cancer. Br J Surg 2014;101: 701–6. 15. George S, Primrose J, Talbot R, et al. Will Rogers revisited: prospective observational study of survival of 3592 patients with colorectal cancer according to number of nodes examined by pathologists. Br J Cancer 2006;95:841–7. 16. O’Shea A, Aly O, Parnaby CN, et al. Increased lymph node yield in colorectal cancer is not necessarily associated with a greater number of lymph node positive cancers. PLoS One 2014;9:e104991. 17. Parsons HM, Tuttle TM, Kuntz KM, et al. Association between lymph node evaluation for colon cancer and node positivity over the past 20 years. JAMA 2011;306:1089–97. 18. Porter GA, Urquhart R, Bu J, et al. Improving nodal harvest in colorectal cancer: so what? Ann Surg Oncol 2012;19:1066–73. 19. Storli K, Sondenaa K, Furnes B, et al. Improved lymph node harvest from resected colon cancer specimens did not cause upstaging from TNM stage II to III. World J Surg 2011;35:2796–803. 20. Moore J, Hyman N, Callas P, et al. Staging error does not explain the relationship between the number of lymph nodes in a colon cancer specimen and survival. Surgery 2010;147:358–65. 21. van Erning FN, Crolla RM, Rutten HJ, et al. No change in lymph node positivity rate despite increased lymph node yield and improved survival in colon cancer. Eur J Cancer 2014;50:3221–9. 22. Sjo OH, Merok MA, Svindland A, et al. Prognostic impact of lymph node harvest and lymph node ratio in patients with colon cancer. Dis Colon Rectum 2012;55:307–15. 23. Bla¨ker H, Hildebrandt B, Riess H, et al. Lymph node count and prognosis in colorectal cancer: the influence of examination quality. Int J Cancer 2015;136:1957–66. 24. Chen L, Kalady MF, Goldblum J, et al. Does reevaluation of colorectal cancers with inadequate nodal yield lead to stage migration or the identification of metastatic lymph nodes? Dis Colon Rectum 2014; 57:432–7. 25. Pihl E, Nairn RC, Milne BJ, et al. Lymphoid hyperplasia: a major prognostic feature in 519 cases of colorectal carcinoma. Am J Pathol 1980;100:469–80. 26. Figueredo A, Coombes ME, Mukherjee S. Adjuvant therapy for completely resected stage II colon cancer. Cochrane Database Syst Rev; 2008:CD005390. 27. Lin HH, Chang YY, Lin JK, et al. The role of adjuvant chemotherapy in stage II colorectal cancer patients. Int J Colorectal Dis 2014;29: 1237–43. 28. Gill S, Loprinzi CL, Sargent DJ, et al. Pooled analysis of fluorouracilbased adjuvant therapy for stage II and III colon cancer: who benefits and by how much? J Clin Oncol 2004;22:1797–806. 29. Johnston PG. Stage II colorectal cancer: to treat or not to treat. Oncologist 2005;10:332–4. 30. Schneider NI, Langner C. Prognostic stratification of colorectal cancer patients: current perspectives. Cancer Manag Res 2014;6:291–300. 31. Zlobec I, Lugli A. Prognostic and predictive factors in colorectal cancer. J Clin Pathol 2008;61:561–9. 32. Pox C, Aretz S, Bischoff SC, et al. S3-Leitlinie Kolorektales Karzinom, Langversion 1.0. Available at: http://leitlinienprogrammonkologie.de/leitlinien.7.0.html. Accessed November 23, 2014. 33. van de Velde CJ, Boelens PG, Borras JM, et al. EURECCA colorectal: multidisciplinary management: European consensus conference colon & rectum. Eur J Cancer 2014;50:1.e1–1.e34.
Acknowledgment The authors are thankful to Kathrin Ferstl-Blahetek and Elfriede Schwarz for excellent technical assistance.
References 1. Chang GJ, Rodriguez-Bigas MA, Skibber JM, et al. Lymph node evaluation and survival after curative resection of colon cancer: systematic review. J Natl Cancer Inst 2007;99:433–41. 2. Edler D, Ohrling K, Hallstrom M, et al. The number of analyzed lymph nodesda prognostic factor in colorectal cancer. Acta Oncol 2007;46:975–81. 3. Goldstein NS. Lymph node recoveries from 2427 pT3 colorectal resection specimens spanning 45 years: recommendations for a minimum number of recovered lymph nodes based on predictive probabilities. Am J Surg Pathol 2002;26:179–89. 4. Swanson RS, Compton CC, Stewart AK, et al. The prognosis of T3N0 colon cancer is dependent on the number of lymph nodes examined. Ann Surg Oncol 2003;10:65–71. 5. Lykke J, Roikjaer O, Jess P, Danish Colorectal Cancer Group. The relation between lymph node status and survival in Stage I-III colon cancer: results from a prospective nationwide cohort study. Colorectal Dis 2013;15:559–65. 6. Feinstein AR, Sosin DM, Wells CK. The Will Rogers phenomenond stage migration and new diagnostic techniques as a source of misleading statistics for survival in cancer. N Engl J Med 1985;312:1604–8. 7. Schemper M, Smith TL. A note on quantifying follow-up in studies of failure time. Control Clin Trials 1996;17:343–6. 8. Resch A, Langner C. Lymph node staging in colorectal cancer: old controversies and recent advances. World J Gastroenterol 2013;19:8515–26. 9. Ma¨rkl B, Kerwel TG, Ja¨hnig HG, et al. Methylene blue-assisted lymph node dissection in colon specimens: a prospective, randomized study. Am J Clin Pathol 2008;130:913–9. 10. Ma¨rkl B, Schaller T, Krammer I, et al. Methylene blue-assisted lymph node dissection technique is not associated with an increased detection of lymph node metastases in colorectal cancer. Mod Pathol 2013;26:1246–54. 11. Ma¨rkl B, Ro¨ssle J, Arnholdt HM, et al. The clinical significance of lymph node size in colon cancer. Mod Pathol 2012;25:1413–22. 12. Simunovic M, Baxter NN. Lymph node counts in colon cancer surgery: lessons for users of quality indicators. JAMA 2007;298: 2194–5. 13. Wong SL, Ji H, Hollenbeck BK, et al. Hospital lymph node examination rates and survival after resection for colon cancer. JAMA 2007; 298:2149–54.