World J Surg (2010) 34:768–775 DOI 10.1007/s00268-009-0336-4
Lymph Node Ratio Versus Number of Affected Lymph Nodes as Predictors of Survival for Resected Pancreatic Adenocarcinoma Imran Bhatti • Oliver Peacock • Altaf K. Awan • David Semeraro • Michael Larvin • Richard I. Hall
Published online: 6 January 2010 Ó Socie´te´ Internationale de Chirurgie 2010
Abstract Background The objective of this study was to compare the prognostic significance of the lymph node ratio (LNR) with the absolute number of affected lymph nodes for resected pancreatic ductal adenocarcinoma. Methods Data were collected from 84 patients who had undergone pancreatoduodenectomy for pancreatic ductal adenocarcinoma over a 10-year period. Patients were categorized into four groups according to the absolute LNR (0, 0–0.199, 0.2–0.299, C0.3). Kaplan-Meier and Cox proportional hazard models were used to evaluate the prognostic effect. Results An LNR of C0.2 (median survival 8.1 vs. 35.7 months with LNR \ 0.2; p \ 0.001) and C0.3 (median survival 5.9 vs. 29.6 months with LNR \ 0.3; p \ 0.001), tumor size (p \ 0.017), positive resection margin (p \ 0.001), and nodal involvement (p \ 0.001) were found to be significant prognostic markers following univariate analysis. Following multivariate analysis, only LNR at both levels [C0.2 (p = 0.05; HR 1.8) and LNR of C0.3 (p = 0.01; HR 2.7)] were independent predictors
I. Bhatti (&) M. Larvin Division of Surgery, School of Graduate Entry Medicine and Health, University of Nottingham Medical School at Derby, Derby City General Hospital, Uttoxeter Road, Derby DE22 3DT, UK e-mail:
[email protected] O. Peacock A. K. Awan R. I. Hall Department of Surgery, Derby City General Hospital, Uttoxeter Road, Derby DE22 3DT, UK D. Semeraro Department of Pathology, Derby City General Hospital, Uttoxeter Road, Derby DE22 3DT, UK
123
of a poor outcome. The number of lymph nodes examined had no effect on overall survival in either node-positive patients (p = 0.339) or node-negative patients (p = 0.473). Conclusions The LNR represents a stronger independent prognostic indicator than the absolute number of affected lymph nodes in patients with resected pancreatic ductal adenocarcinoma.
Introduction Pancreatic cancer is associated with an extremely poor prognosis. At best, only one-fifth of individuals present with operable disease [1]. Following resection, the prognosis remains poor with median and 5-year survival times of 12 to 18 months [2] and 15% to 20% [3], respectively. The outcome of resected pancreatic cancer has been associated with several histologic factors [4–9] including tumor size [10] and differentiation [10] and the resection margin status [11]. More recently, a relation has been demonstrated between lymph node involvement and poor survival [5]. However, this association has not always been consistent [12], so its use as a prognostic marker has been limited. Several studies have assessed the relation of lymph node ratio (number of metastatic lymph nodes divided by the number of examined nodes) (LNR) in contrast to the presence of nodal disease to identify a more significant and consistent prognostic indicator [5, 6, 13–15]. From these data, LNR has been found to be the most reliable factor influencing outcome. Using LNR as a categoric variable, previous studies have shown a cutoff point between 0.15 and 0.30 to have prognostic significance [13, 14]. The objective of this study was to compare the prognostic significance of the LNR with the absolute number of
World J Surg (2010) 34:768–775
affected lymph nodes for resected pancreatic ductal adenocarcinoma.
Methods Patients and techniques Between 1998 and 2008, restrospective data were collected from 84 patients undergoing classic Kausch-Whipple resection with standard lymphadenectomy. Only patients with histologic confirmation of pancreatic ductal adenocarcinoma were included in the study. Patient demographics, lymph node involvement with total resection, operative details, intervention for biliary decompression, and standard histologic tumor characteristics were recorded. The operative procedure used was pancreatoduodenectomy and reconstruction by forming a pancreatojejunostomy with a single-layer, full-thickness, end-to-side anastomosis (pancreatic parenchyma to jejunum). All patients underwent standard lymphadenectomy, which included resection of nodes within the confines of the hepatoduodenal ligament, right side of the superior mesenteric artery, and inferior vena cava. Adjuvant chemotherapy was offered to all patients following recovery from the operative procedure provided there were no known contraindications to chemotherapy drugs (i.e., bone marrow suppression or hypersensitivities). None of the patients received neoadjuvant chemotherapy or radiotherapy. The LNR was calculated for the 84 patients. Patients were then subdivided into four groups according to their LNR value (0, 0–0.199, 0.2–0.299, C0.3). Theses cutoff points were used as they included the range of LNRs found to have a significant relation with survival in previous studies. The resected specimen was immediately fixed in formalin. Surgical margins were marked with black ink prior to specimen handling. Following dissection and sampling, tissue blocks were processed and subsequently embedded in paraffin. After processing and sectioning, peripancreatic lymph nodes were stained with hematoxylin and eosin (H&E) prior to evaluation. All pancreatic cancer specimens were processed and reported according to the Royal College of Pathologists’ guidelines (UICC TNM classification, 6th edition) that define a positive margin to include at least one cancer cell within 1 mm of any resection margin. Specimens were examined and reported by experienced pathologists in all cases, including documentation of total number of lymph nodes (LNs) yielded and the number of nodes containing metastatic tumor.
769
To verify pathologic findings, the specimens were reexamined by another senior pathologist who was unaware of the initial report. If there were any discrepancies in their findings, a third independent pathologist examined the specimen. The results that were reported the same by two pathologists were used in the study. Both hospital records and general practitioner databases were used to identify the length of survival accurately. Patients surviving less than a month were excluded from the study. All patients were followed up initially at 6 weeks with a history, physical examination, and routine blood tests. After the initial assessment, patients were seen in the clinic every 6 months with a yearly computed tomography (CT) scan to identify recurrence. No patients were lost to follow-up. Statistical analysis Continuous data were analyzed using the median, interquartile range (IQR), and 95% confidence intervals (CI). The v2 or Fisher’s exact test was used for comparative analysis of categoric data. Survival data were analyzed using log-rank testing for the univariate analysis and the Cox proportional hazards for the multivariate analysis. Corrected log-rank p values were used for univariate survival whereas continuous prognostic data were dichotomized. All patients who died within 30 days of surgery were excluded from survival analysis, and p B 0.05 was taken to reflect significance. Statistical analysis was performed using an SPSS database (SPSS for Windows, version 16.0; SPSS, Chicago, IL, USA).
Results A total of 106 patients underwent pancreatoduodenectomy with standard lymphadenectomy for pancreatic ductal adenocarcinoma during the study period. Perioperative mortality was 3.8%, and another 18 patients did not have complete follow-up data, resulting in the data of 84 patients used for survival analysis. Data regarding the number of lymph nodes involved and resected were available for all 84 patients. Demographic and lymph node data are shown in Table 1. A disease-free margin was achieved in 58% of patients, and the median number of nodes examined was 9 (range 1– 26). Among the 84 patients, 56 (67%) had nodal disease; 20 (24%) had one diseased LN, and 36 (42%) had more than one LN involved. The other 28 (33%) patients were free of nodal disease. The median number on lymph nodes involved was 1 (0–9), and the median LNR was 0.16 (0– 1.00) (Table 1). LNR was stratified into the following groups: \0.2, 0.2–0.29, and C0.30.
123
770
World J Surg (2010) 34:768–775
Table 1 Demographics and lymph node status for patients with a resected pancreatic ductal adenocarcinoma Demographics No. of patients identified
84
Sex (M:F)
48:36
Median age (IQR)
65 (51–79) years
Overall mean survival (95% CI)
22 (17–28) months
Preoperative stenting (n, %) Yes
68 (81%)
No
16 (19%)
Adjuvant chemotherapy (n, %) Yes
30 (36%)
No
54 (64%)
No. of jaundice cases preop (Bili [35 lmol/L) (n, %) Present
23 (27%)
Absent
61 (73%)
Resection margin (n, %) Negative Positive
49 (58%) 35 (42%)
Recurrence (n, %) Locoregional
18 (21%)
Distant
30 (36%)
Nodal status (n, %) Positive
56 (67%)
Negative
28 (33%)
Involved nodes (median, range)
1 (0–9)
Examined nodes (median, range)
9 (1–26)
Examined nodes (mean, mode)
9.17 (8)
Lymph node ratio (median, range)
0.16 (0–1.00)
Multivariate analysis
Lymph node ratio (n, %) 0
27 (32%)
[0–0.199 0.2–0.299
19 (23%) 12 (14%)
C0.3
26 (31%)
IQR interquartile range, CI confidence interva, preop preoperative, Bili bilirubin
To date, 59 of the 84 patients have died (median survival 10.5 months). The 3- and 5-year survival rates were 24% and 13%, respectively. The overall median survival was 22.0 months. Univariate analysis The univariate analysis shown in Table 2 confirms that tumor size (p = 0.017), resection margin (p \ 0.001), lymph node involvement (p \ 0.001) (Fig. 1), and LNR (p \ 0.001) (Fig 2) were pathologic parameters that significantly influenced survival.
123
The worst survival was identified in patients with an LNR of 0.2 to 0.29 [8.1 months, 95% confidence interval (CI) 7.6–8.5] and C0.3 (5.9 months; 95% CI 0.5–11.3) compared with an LNR of 0 (46.3 months, 95% CI 26.5– 66.1) and 0 to 0.199 (26.8 months, 95% CI 10.3–43.3) (p \ 0.001) (Table 2; Fig 2). Therefore, the optimum stratification points that demonstrated the most survival difference were taken at 0.2 (LN ratio group 1) and 0.3 (LN ratio group 2) (p \ 0.001) (Table 2; Figs. 3, 4). These cutoff points were consistent with those in similar studies investigating the role of LNR as a possible prognostic factor [13, 14, 16]. In our cohort, both LN involvement and LNR had a clear relation with survival. However in other studies, LN involvement alone did not predict survival, as patients with one affected LN had the same survival as patients with uninvolved LNs [12, 17]. The univariate analysis revealed that patients with an LNR [0.3 clearly had the worst outcome, with a median survival of 5.9 months (Table 2). Certain parameters—number of LNs examined (cutoff 15), sex, age, adjuvant chemotherapy, preoperative biliary stenting, and tumor differentiation—did not influence survival (Table 2). Following subgroup analysis of patients with involved LNs, LNR continued to influence survival (Table 2, subgroup analysis). Furthermore, on examining the whole group, the number of examined LNs (B15 vs. [15 LNs) had no influence on survival for both the subgroup with positive LNs (p = 0.339) (Table 2) and the subgroup with negative LNs (p = 0.473) (Table 2).
Multivariate analysis was undertaken using two different states (state 1, LNR cutoff 0.2; and state 2, LNR cutoff 0.3) (Table 3). LNR remained a significant predictor of survival at both cutoff points (0.2 and 0.3) using multivariate analysis, whereas tumor size, resection margin, and LN involvement lost significance. An LNR of [0.3 was again shown to be the strongest predictor for outcome (p = 0.013; hazards ratio 2.7).
Discussion Surgical resection of pancreatic adenocarcinoma for cure remains a major challenge. Numerous studies have demonstrated that the presence of nodal disease is an important predictor of survival [3, 8, 18–20]. Despite this, it is important to note that nodal disease alone did not predict survival in some of the larger studies [12]. In our cohort, nodal disease did predict survival with univariate analysis
World J Surg (2010) 34:768–775 Table 2 Univariate survival analysis of resected pancreatic adenocarcinoma
771
Parameter
No of patients
Median survival (95% CI)
p (log-rank)
Tumor size 6
34.0 (1.0–100.6)
[20 mm without invasion
B20 mm
20
26.8 (5.6–48.0)
[20 mm with invasion
58
10.5 (8.9–12.1)
Negative
49
23.2 (9.3–37.0)
Positive
35
7.1 (3.6–10.7)
0.017
Resection margin status \0.001
Tumor differentiation Moderate
60
13.5 (11.3–15.7)
Poor
24
10.0 (5.2–14.8)
LNR 0
0.947
\0.001
27
46.3 (26.5–66.1)
[0–0.199
19
26.8 (10.3–43.3)
0.2–0.299
12
8.1 (7.6–8.5)
C0.3
26
5.9 (0.5–11.3)
\0.2
46
35.7 (15.0–56.5)
C0.2
38
8.1 (6.1–10.1)
\0.3
58
29.6 (12.1–47.1)
C0.3
26
5.9 (0.5–11.3)
Negative
29
30.7 (25.5–36.0)
Positive
55
9.8 (7.8–12.0)
30 54
14.6 (1.0–31.0) 10.3 (6.3–14.3)
0.258
No
16
15.2 (8.8–16.1)
0.561
Yes
68
12.5 (3.6–10.7)
B60 Years
25
15.4 (10.9–19.9)
[60 Years
59
14.4 (9.8–19.1)
Male
48
13.5 (12.0–15.0)
Female
36
26.8 (11.6–42.0)
B15
35
10.8 (6.6–15.1)
[15
20
13.1 (6.2–20.1)
14 15
35.7 (25.7–45.7) 64.0 (26.8–101-2)
0.473
\0.2
17
26.8 (11.5–42.1)
0.001
C0.2
38
7.6 (6.1–10.1)
\0.3
29
14.6 (11.5–17.7)
C0.3
26
5.9 (0.50–11.3)
LN ratio group 1 \0.001
LN ratio group 2 \0.001
Nodal involvement \0.001
Adjuvant chemotherapy Yes No Preoperative stent
Age 0.528
Sex 0.817
Subgroup analysis No. of examined LNs in node-positive patients 0.339
No. of examined LNs in node-negative patients B15 [15 LNR in node-positive patients
LNR in node-negative patients LNR lymph node ratio, LN lymph node
0.002
123
772
World J Surg (2010) 34:768–775
Fig. 1 Kaplan–Meier survival curve for pancreatic ductal adenocarcinoma according to lymph node (LN) involvement
Fig. 2 Kaplan–Meier survival curve for pancreatic ductal adenocarcinoma according to different cutoff points of the lymph node ratio (LNR_Cat). Cat refers to categories
but failed to show an association with survival with multivariate analysis. Using the presence of nodal disease or absolute number of affected lymph nodes may introduce bias due to the inevitable possibility of incomplete lymphadenectomy or inadequate histopathologic examination [6, 13, 14]. To
123
overcome the potential problems associated with nodal staging mentioned above and to define better the prognostic role of nodal disease, several studies have used the LNR as a prognostic parameter for various gastrointestinal tumors [21–23]. Recent studies have looked into the role of LNR in predicting survival of patients with pancreatic ductal
World J Surg (2010) 34:768–775
773
Fig. 3 Kaplan–Meier survival curve for pancreatic ductal adenocarcinoma according to the LNR cutoff of 0.2
Fig. 4 Kaplan–Meier survival curve for pancreatic ductal adenocarcinoma patients according to the LNR cutoff of 0.3
adenocarcinoma [5, 6, 13, 14]. These studies suggested future use of LNR in the stratification of prognosis with an ideal cutoff point at 0.15 to 0.30.
Our study was also able to demonstrate that an LNR cutoff of 0.3 is the most potent prognostic factor based on the multivariate analysis. The cutoff values shown in
123
774
World J Surg (2010) 34:768–775
Table 3 Multivariate survival analysis (Cox regression) Parameter
p
Hazards ratio (95% CI)
State 1 Tumor size
0.73
0.9 (0.2–2.8)
Positive margin
0.70
0.8 (0.4–1.9)
LN involved
0.20
0.6 (0.3–1.3)
LNR C0.2
0.05
1.8 (1.1–2.2)
State 2 Tumor size
0.56
0.7 (0.2–2.4)
Positive margin
0.27
0.6 (0.3–1.4)
LN involved
0.06
2.0 (1.3–3.7)
LNR C0.3
0.01
2.7 (1.6–4.4)
Conclusions
State 1: LNR cutoff 0.2; State 2: LNR cutoff 0.3
Table 4 Distribution of histologic tumor characteristics stratified by LNR Characteristic
No. of patients stratified by LNR \0.3
C0.3
B20 mm
5
1
[20 mm
53
25
of C3 (p = 0.002 and p = 0.003, respectively) (Table 4). Therefore, the LNR may predict both resectability and distant recurrence. It maybe argued that assessment of a small number of lymph nodes may result in understaging disease by missing metastatic nodes. However, no correlation could be made between the number of nodes examined (in both nodepositive and node-negative patients) and survival. Therefore, patients in this series did not appear to be understaged.
p*
The LNR is the strongest predictor for survival in patients following resection of pancreatic ductal adenocarcinoma. This study confirms the findings of other centers with the addition that LNR predicts a positive resection margin status and distant recurrence of disease. LNR is an important piece of information that could be used for assessing prognosis and for selecting the specific type(s) of adjuvant treatment that best treats aggressive disease.
Tumor size 0.661
Acknowledgment The authors appreciate the statistical guidance from Dr. Graham Warren, PhD, Department of Medical Statistics, University of Nottingham, UK.
Resection margin Negative
37
4
Positive Recurrence
10
9
Locoregional
14
4
3
27
No
36
18
Yes
22
8
No
44
24
Yes
14
2
Distant
0.002
References 0.003
Adjuvant therapy 0.626
Preoperative stent 0.131
* Fisher’s exact tests calculated were for 3 9 2 contingency tables
previous studies were comparable (0.15–0.30). Furthermore, LNR continued to be a significant predictor for survival following subgroup analysis in patients with positive lymph nodes. Previous randomized control trials have shown improved survival following adjuvant chemotherapy. In our series, neither adjuvant chemotherapy (p = 0.258) (Tables 2, 4) nor preoperative biliary stenting (p = 0.561) (Tables 2, 4) were associated with length of survival using an LNR cutoff of 0.3. However it should be noted that there was a significantly greater proportion of patients with a positive resection margin and distant recurrence in the group with an LNR
123
1. James TA, Sheldon DG, Rajput A et al (2004) Risk factors associated with earlier age of onset in familial pancreatic carcinoma. Cancer 101:2722–2726 2. Alexakis N, Halloran C, Raraty M et al (2004) Current standards of surgery for pancreatic cancer. Br J Surg 91:1410–1427 3. Richter A, Niedergethmann M, Sturm JW et al (2003) Long-term results of partial pancreaticoduodenectomy for ductal adenocarcinoma of the pancreatic head: 25-year experience. World J Surg 27:324–329 4. Connor S, Bosonnet L, Ghaneh P et al (2004) Survival of patients with periampullary carcinoma is predicted by lymph node 8a but not by lymph node 16b1 status. Br J Surg 91:1592–1599 5. Sierzega M, Popiela T, Kulig J et al (2006) The ratio of metastatic/resected lymph nodes is an independent prognostic factor in patients with node-positive pancreatic head cancer. Pancreas 33:240–245 6. Berger AC, Watson JC, Ross EA et al (2004) The metastatic/ examined lymph node ratio is an important prognostic factor after pancreaticoduodenectomy for pancreatic adenocarcinoma. Am Surg 70:235–240; discussion 240 7. Neoptolemos JP, Stocken DD, Dunn JA et al (2001) Influence of resection margins on survival for patients with pancreatic cancer treated by adjuvant chemoradiation and/or chemotherapy in the ESPAC-1 randomized controlled trial. Ann Surg 234:758–768 8. Winter JM, Cameron JL, Campbell KA et al (2006) 1423 Pancreaticoduodenectomies for pancreatic cancer: a single-institution experience. J Gastrointest Surg 10:1199–1210; discussion 1210– 1211 9. Bassi C, Stocken DD, Olah A et al (2005) Influence of surgical resection and post-operative complications on survival following
World J Surg (2010) 34:768–775
10. 11.
12.
13.
14.
15.
16.
adjuvant treatment for pancreatic cancer in the ESPAC-1 randomized controlled trial. Dig Surg 22:353–363 Ueda M, Endo I, Nakashima M et al (2009) Prognostic factors after resection of pancreatic cancer. World J Surg 33:104–110 Kure S, Kaneko T, Takeda S et al (2005) Analysis of long-term survivors after surgical resection for invasive pancreatic cancer. HPB (Oxford) 7:129–134 Sohn TA, Yeo CJ, Cameron JL et al (2000) Resected adenocarcinoma of the pancreas-616 patients: results, outcomes, and prognostic indicators. J Gastrointest Surg 4:567–579 House MG, Gonen M, Jarnagin WR et al (2007) Prognostic significance of pathologic nodal status in patients with resected pancreatic cancer. J Gastrointest Surg 11:1549–1555 Pawlik TM, Gleisner AL, Cameron JL et al (2007) Prognostic relevance of lymph node ratio following pancreaticoduodenectomy for pancreatic cancer. Surgery 141:610–618 Tamandl D, Kaczirek K, Gruenberger B et al (2009) Lymph node ratio after curative surgery for intrahepatic cholangiocarcinoma. Br J Surg 96:919–925 Kulig J, Sierzega M, Kolodziejczyk P et al (2009) Ratio of metastatic to resected lymph nodes for prediction of survival in patients with inadequately staged gastric cancer. Br J Surg 96:910–918
775 17. Riediger H, Keck T, Wellner U et al (2009) The lymph node ratio is the strongest prognostic factor after resection of pancreatic cancer. J Gastrointest Surg 13:1337–1344 18. Brennan MF, Kattan MW, Klimstra D et al (2004) Prognostic nomogram for patients undergoing resection for adenocarcinoma of the pancreas. Ann Surg 240:293–298 19. Lim JE, Chien MW, Earle CC (2003) Prognostic factors following curative resection for pancreatic adenocarcinoma: a population-based, linked database analysis of 396 patients. Ann Surg 237:74–85 20. Breslin TM, Hess KR, Harbison DB et al (2001) Neoadjuvant chemoradiotherapy for adenocarcinoma of the pancreas: treatment variables and survival duration. Ann Surg Oncol 8:123–132 21. Bogoevski D, Onken F, Koenig A et al (2008) Is it time for a new TNM classification in esophageal carcinoma? Ann Surg 247:633– 641 22. Sierra A, Regueira FM, Hernandez-Lizoain JL et al (2003) Role of the extended lymphadenectomy in gastric cancer surgery: experience in a single institution. Ann Surg Oncol 10:219–226 23. Rosenberg R, Friederichs J, Schuster T et al (2008) Prognosis of patients with colorectal cancer is associated with lymph node ratio: a single-center analysis of 3, 026 patients over a 25-year time period. Ann Surg 248:968–978
123 View publication stats