Very High Serum CA 19-9 Levels: A Contraindication to Pancreaticoduodenectomy?

J Gastrointest Surg (2009) 13:1791–1797 DOI 10.1007/s11605-009-0916-5

2008 SSAT POSTER PRESENTATION

Very High Serum CA 19-9 Levels: A Contraindication to Pancreaticoduodenectomy? O. Turrini & C. M. Schmidt & J. Moreno & P. Parikh & J. M. Matos & M. G. House & N. J. Zyromski & A. Nakeeb & H. A. Pitt & K. D. Lillemoe

Received: 25 February 2009 / Accepted: 15 April 2009 / Published online: 21 May 2009 # 2009 The Society for Surgery of the Alimentary Tract

Abstract Aim To assess the outcome of patients with resectable pancreatic adenocarcinoma (PA) associated with high serum CA 19-9 levels. Methods From 2000 to 2007, 344 patients underwent pancreatoduodenectomy for PA. Fifty-three patients (elevated group) had preoperatively elevated serum CA 19-9 levels (>400 IU/ml) after resolution of obstructive jaundice. Of these, 27 patients had high levels (400–899 IU/ml (HL)) and 26 patients had very high levels ≥900 IU/ml (VHL). Fifty patients with normal preoperative serum CA 19-9 levels (37 IU/ml and 400 IU/ml were designated the elevated group. Patients were also subcategorized as having a high level (HL) if the serum CA19-9 level was 400 IU/ml to 899 IU/ml or a very high level (VHL) if the serum CA19-9 level was >900 IU/ml. During the same period, patients with normal preoperative serum CA 19-9 levels (≤37 IU/ml) comprised our control group. The 400 IU/ ml cut-off level was chosen because based upon recent literature this level was at or above majority the level considered to dictate a poor prognostic outcome (Table 1). By using this 400 IU/ml cut-off, we sought to compare the survival of patients anticipated to have the worst prognosis with the survival of patients anticipated to have the best prognosis according to preoperative serum CA19-9 level. Surgery All patients underwent PD with curative intent. Laparoscopy was not performed routinely; however, one patient of control group and three patients of elevated group had laparoscopy prior to PD. End Points Studied The variables evaluated included age, sex, weight loss, pre- and postoperative serum CA 19-9 level (from 1 to 3 months after surgery and before any adjuvant treatment), maximal tumor size (cm) defined as maximum diameter at pathologic analysis, histologic differentiation (well, moderate, or poor), margin of resection (positive or negative), node stage (positive nodes; number of examined nodes), metastasis stage, and perineural, vascular, and lymphatic invasion. Margins assessed included the pancreatic neck, bile duct, uncinate/retroperitoneal, and duodenal. Statistical Analysis Data analyses were carried out with GraphPad Prism (GraphPad Software Inc., San Diego, CA, USA) and Excel 2004 (Microsoft, Seattle, WA, USA). Survival time was measured from the time of PD until death

Year

N

Cut-off (IU/ml)

Ca19-9>1,000 (n)

2008 2008 2008 2008 2008 2005 2004 1998 1997

385 104 52 94 53 63 129 148 40

180a 353 150 150 473 100 200 2,000 180a

None None None None NA None NA yes (>15) None

J Gastrointest Surg (2009) 13:1791–1797

1793

or last follow-up (censor date was November 1, 2008). Statistical associations between categorical factors were assessed using the Fisher exact test. The association of categorical factors with survival was assessed using the Kaplan–Meier method and was tested using the log-rank test. Statistical significance was set at p value 400 IU/ml – Elevated group (n=53)

Results CA19-9 >899 IU/ml VHL group (n=26)

Of the 344 consecutive patients with resectable PA who underwent PD, 286 patients matched the inclusion criteria for this study. Fifty patients met criteria for the control group (CA 19-9≤37 IU/ml). Fifty-three patients met criteria for the elevated group (CA19-9≥400 IU/ml) after matching the exclusion criteria. Within the elevated group, HL (CA19-9 400–899 IU/ml) and VHL (CA 19-9≥900 IU/ml) subgroups comprised 27 patients and 26 patients, respectively (Fig. 1).

CA19-9 still elevated (n=20)

Clinical Characteristics Clinical and pathological characteristics of the elevated and control groups were comparable (Table 2). No patients were lost to follow-up and median follow-up was 47 months (95% CI [45.4–56.3]). Overall perioperative morbidity and mortality for all 103 patients

CA19-9 400-899 IU/ml HL group (n=27)

Postop CA19-9 (n=31)

CA 19-9 normalization (n=11) (HL, n=8; VHL, n=3)

Figure 1 Selection of patients.

Table 2 Comparison of Clinical and Pathologic Parameters in Control (CA 19-9≤37 IU/ml) and Elevated (CA 19-9≥400 IU/ ml) Serum CA 19-9 Groups

Results are shown as median (range) except follow-up which is expressed as median with 95% confidence interval ([CI]). Other parameters are expressed as n (%) where n=number of patients and %=percentage of patients LOS length of hospital stay

Control group (n=50)

Elevated group (n=53)

p value

Age Male n (%) Weight loss n (%) Preoperative biliary stenting n (%) Preoperative CA 19-9 (IU/ml) Median follow-up (months) Operative duration (min) Blood loss (ml) Vascular resection n (%) Tumor size (cm)

64 (45–82) 19 (38) 28 (56) 14 (28) 33 (1–36) 42 ([37.6–53.5]) 344 (182–561) 724 (150–5,000) 12 (24) 3 (range 0.6–4.5)

66 (44–80) 21(40) 32 (60) 23 (43) 1,756 (400–13,100) 51 ([47.9–63]) 324 (190–697) 1,082 (300–4,000) 14 (26) 3 (range 0.9–5.9)

Ns Ns Ns Ns 0.02 Ns Ns Ns Ns Ns

Tumor differentiation n (%) Poor Moderate Well Positive margin n (%) Examined lymph nodes N1 status n (%) Perineural invasion n (%) Perivascular invasion n (%) Morbidity n (%) Mortality n (%) LOS (days) Adjuvant treatment n (%)

19 (38) 24 (48) 7 (14) 9 (18) 12 (5–29) 29 (58) 20 (40) 19 (38) 18 (36) 0 13 (range 6–68) 28 (56)

29 (55) 21 (40) 3 (5) 13 (24) 12 (4–28) 39 (73) 29 (55) 27 (51) 15 (28) 0 11 (range 6–27) 34 (64)

Ns Ns Ns Ns Ns Ns Ns Ns Ns Ns Ns Ns

1794

Patients in italics are patients with preoperative VHL (>900 IU/ml) a

Alive without recurrence

b

Death

Patients

Preoperative CA19-9 serum level (IU/ml)

Postoperative CA19-9 serum level (IU/ml)

Survival (months)

1 2

408 467

27 30

11b 17b

3 4 5 6 7 8 9 10 11

543 560 564 569 571 847 992 4,649 13,100

15 33 17 25 20 34 31 18 16

23b 29b 20b 26b 31b 6b 62b 82a 12b

was 32% and 0%, respectively. Postoperative serum CA 199 level was available in 31 patients of the elevated group (58%). Postoperative serum CA 19-9 level normalization occurred in 11 patients (eight HL, three VHL; 21%) (Table 3). Conversely, 20 patients (38%) had postoperative CA19-9 serum level decreasing without reaching the normal range or increasing compared to their preoperative value. Survival Analysis The median overall survival of the control group (n=50) versus the elevated group (n=53) was 22 vs. 15 months, respectively (p=0.02). Overall 3-year overall survival was 32% in the control group vs. 14% in the elevated group, respectively (p=0.03) (Fig. 2). Within the elevated group, there was no statistical difference in median survival (15 and 12 months) and 3-year overall survival (13% and 15%) between patients with HL or VHL (Fig. 3). The median overall survival of patients who normalized serum CA 19-9 level post-resection (n=11) or not (n=20) was 23 and 16 months, respectively (p=0.02) (Fig. 4). There was no statistical difference in median (23 and 22 months) and 3-year overall survival (32% and 27%) between patients who normalized their serum CA 19-9 level and patients in the control group (Fig. 4).

Discussion

benefit from surgery remains a major challenge in the current care of patients with pancreatic cancer. The serum CA 19-9 tumor antigen is currently the most clinically useful serologic marker for pancreatic cancer. The majority of PA will secrete CA 19-9 and have measurable serum levels.20 Thus, many investigators have turned to serum CA 19-9 as a possible prognostic marker for tumor

100

control group elevated group 75

survival probability (%)

Table 3 Patients with Normalized (CA 19-9≤37 IU/ml) Postoperative Serum CA 19-9 Level

J Gastrointest Surg (2009) 13:1791–1797

50

25

0 0

12

24

36

Time (months)

The radiographic ability to identify metastatic or locally unresectable PA continues to improve. Nonetheless, even with careful preoperative staging using state-of-the-art technology, the prevalence of undetectable metastatic or locally advanced disease remains approximately 15– 20%.18The median survival for patients with unresectable disease is 6–12 months and only systemic therapies have demonstrated a potential survival benefit.19 Sparing patients with unresectable disease who will not obtain a survival

(log-rank test p=0.03) Patients alive at risk Time (months)

0

12

24

Control group

50

39

16

36 5

Elevated group

53

31

11

5

Figure 2 Survival of patients having pancreatoduodenectomy and with preoperative normal (CA19-9≤37 IU/ml) (n=50) or elevated (CA 19-9≥400 IU/ml) (n=53) CA19-9 serum level.

J Gastrointest Surg (2009) 13:1791–1797

1795

patients with HL survival probability (%)

75

patients with VHL

50

25

0 0

12

24

36

Time (months)

(log-rank test p=0.95) Patients alive at risk Time (months)

0

12

24

36

HL group

27

13

7

3

VHL group

26

18

4

4

Figure 3 Survival of patients having pancreatoduodenectomy and with preoperative high level (HL, CA 19-9 400–899 IU/ml) (n=27) or very high level (VHL, CA19-9>899 IU/ml) (n=26) serum CA 19-9.

resectability, recurrence, and patient survival. Importantly, CA 19-9 levels detected by conventional antibody tests may be affected by Lewis blood group phenotypes.20 In fact, pancreatic cancer patients with a Lewis negative (a-, b-) phenotype will have an undetectable CA 19-9 level. Moreover, 7% to 10% of patients may have undetectable CA 19-9 levels even in the face of metastatic or recurrent disease.21 Berger et al. showed that patients with undetectable CA 19-9 levels actually had improved survival.16 The findings of our study are not statistically different by including or excluding patients with undetectable levels which reinforces our findings that normalization of preoperatively elevated serum CA 19-9 is a good prognostic sign and confers a survival advantage. Corroborating our study in part is Ferrone et al.9 who found that a postoperative decrease in serum CA 19-9 level and an absolute postoperative serum CA 19-9 value less than 200 IU/ml were both significant predictors of survival in patients with PA. Another use of preoperative serum CA 19-9 level may be in its ability to detect patients at greater risk for having

undetected occult metastatic or locally advanced disease.12 Moreover, some investigators have found that elevated preoperative serum CA 19-9 levels are significantly associated with tumor unresectability, although the cut-off levels reported range from 100 to 350 IU/ml.8–17 Based on these data, some speculate that preoperative staging of a potentially resectable PA should routinely include serum CA 19-9 levels after biliary decompression. In the case where careful preoperative staging indicates a radiographically resectable PA without evidence of distant metastasis but the serum CA 19-9 levels are highly elevated, the staging might be better clarified by the use of laparoscopy. Indeed, the benefit of laparoscopy in the radiographically resectable patient with normal CA 19-9 is still under debate. However, laparoscopy has a higher yield in detecting unknown carcinomatosis or liver metastasis in 5% to 10% of patients if performed in patients with high serum CA 19-9 levels.4, 5, 8 Alternatively, a neoadjuvant approach would allow time and follow-up restaging which may spare resection in patients where progressive disease is imminent. The time selection of the neoadjuvant approach, however, is also not uniformly reliable in weeding out micrometastatic and early recurrence patients.22 normalized CA 19-9

100

control group still elevated CA 19-9 75

survival probability (%)

100

50

25

0 0

12

24

36

Time (months) (log rank test normalized vs. control

p=0.78

log rank test normalized vs. still elevated

p=0.02)

Patients alive at risk Time (months)

0

12

24

36

normalized group

11

9

6

3

still elevated group

20

14

6

3

control group

50

39

16

5

Figure 4 Survival of patients in the elevated group who normalized (CA19-9≤37 IU/ml) (n=11) serum CA 19-9 level post-resection are compared to patients who failed to normalize (n=20) post-resection and patients of the control group (i.e., normal preoperative CA 19-9 serum level).

1796

Our study confirmed that elevated serum CA 19-9 level correlates with a poor survival as corroborated by several studies.8–17, 23 Our study also found that patients with VHL (over 800 IU/ml) had equivalent survival as patients with HL. Thus, although a serum CA 19-9 level of 150 or 300 IU/ml preoperatively may discriminate between patients with good and poor outcome,5, 12–14, 16 levels over 400 IU/ml, however high, have no additional effect on survival. Median survival of patients with ≥400 IU/ml serum CA19-9 levels is higher than patients with metastatic or unresectable disease.18 Thus, patients with resectable PA must be given the benefit of the doubt and be offered resection despite high serum levels of CA 19-9 levels. Consideration may be given in these patients to enrollment in an aggressive adjuvant regimen to improve survival.24, 25 Postoperative changes of serum CA 19-9 levels have been examined previously. Indeed, Montgomery et al.10 demonstrated that normalization of serum CA 19-9 was a good prognostic factor. On the other hand, the cut-off in this study was 180 IU/ml and very few patients with high serum CA 19-9 levels (>400 IU/ml) were enrolled. Nakao et al.17 published over 10 years ago a report of 15 resected patients with serum CA 19-9 levels >2,000 IU/ml. These patients had a median survival of 6 months and were all dead after 19 months of follow-up. Recently, Hernandez et al.26 reported the importance of velocity of normalization in predicting a favorable prognosis. However, Hernandez’s series did not include patients with very high serum CA 199 levels. In our series, normalization of serum CA 19-9 levels was not rare and occurred in 21% of patients. Moreover, the novel aspect of this study is that even patients with preoperative VHL (992, 4,649, and 13,100 IU/ml) may normalize their serum CA 19-9 level and have equivalent overall survival compared to patients with normal preoperative serum CA 19-9 levels. We would speculate that absence of normalization is a marker of persistent tumor burden after PD. Unfortunately, no preoperative criteria permitted us to predict which patients would normalize serum CA 19-9 levels and likely to benefit from surgery. Thus, we propose that patients with highly elevated serum CA 19-9 level (after biliary decompression) have optimal and timely (