Histopathology 2015, 66, 457–462. DOI: 10.1111/his.12577

SHORT REPORT

Human equilibrative nucleoside transporter 1 testing in pancreatic ductal adenocarcinoma: a comparison between murine and rabbit antibodies Magali Svrcek,1,2 J
er^ ome Cros,3,4 Raphael Mar
echal,5 Jean-Baptiste Bachet,2,6 Jean-Francßois Fl
ejou1,2 & Pieter Demetter7 1

Department of Pathology, AP-HP, H^opitaux Universitaires Est Parisien, Saint-Antoine Hospital, Paris, France, Universite Pierre et Marie Curie – Paris 6, Paris, France, 3AP-HP, H^opitaux Universitaires Paris Nord Val de Seine, Beaujon, France, 4Universite Paris Diderot – Paris 7, Paris, France, 5Department of Gastroenterology and Gastrointestinal Cancer Unit, Erasme Hospital, Universite Libre de Bruxelles, Belgium, France, 6Department of HepatoGastroenterology and Digestive Oncology, AP-HP, H^opitaux Universitaires Est Parisien, Pitie Salp^etriere Hospital, Paris, France, and 7Department of Pathology, Erasme Hospital, Universite Libre de Bruxelles, Belgium, France

2

Date of submission 4 June 2014 Accepted for publication 5 October 2014 Published online Article Accepted 9 October 2014

Svrcek M, Cros J, Mar
echal R, Bachet J-B, Fl
ejou J-F & Demetter P (2015) Histopathology 66, 457–462. DOI: 10.1111/his.12577

Human equilibrative nucleoside transporter 1 testing in pancreatic ductal adenocarcinoma: a comparison between murine and rabbit antibodies Aims: The human equilibrative nucleoside transporter 1 (hENT1) expression level in pancreatic ductal adenocarcinoma (PDAC) may predict survival in gemcitabine-treated patients after resection. These results have been obtained with a murine anti-hENT1 antibody (10D7G2) that is not commercially available. Another antibody, which is rabbit-derived (SP120), appears to have no predictive value in local, advanced or metastatic PDAC. We aimed to study whether the two antibodies are equivalent. Methods and results: We compared hENT1 expression with both antibodies in resected PDAC. The results were correlated with overall survival (OS) following gemcitabine treatment. Tissues from two sets of

patients (n = 147 each) were stained with SP120 by the use of different equipment, with an amplification technique being used for set 2. The rate of ‘hENT1 high’ cases was lower with SP120 (set 1, 7% versus 48%; set 2, 11% versus 38%). With the amplification technique, the rate of hENT1 high cases was globally similar between both antibodies. However, concordance between the antibodies was found in only 50% of cases. High hENT1 expression was predictive of OS only with 10D7G2 (hazard ratio 0.49; 95% confidence interval 0.24–0.98; P = 0.045). Conclusions: The two antibodies are not equivalent. Further prospective studies seem to be warranted before hENT1 testing for PDAC is used in daily practise.

Keywords: hENT1, immunohistochemistry, pancreatic adenocarcinoma

Introduction Address for correspondence: M Svrcek, Service d’Anatomie et Cytologie Pathologiques, H^ opital Saint-Antoine, 184, rue du Faubourg Saint-Antoine, F-75571 Paris Cedex 12, France. e-mail: [email protected] M.S. and J.C. contributed equally to this work. © 2014 John Wiley & Sons Ltd.

Gemcitabine is one of the key drugs for adjuvant chemotherapy in patients who undergo surgery for pancreatic ductal adenocarcinoma (PDAC). The potential role of human equilibrative nucleoside transporter 1

458

M Svrcek et al.

Table 1. Patient characteristics

Variables

Set 1 (n = 147)

Table 1. (Continued) Set 2 (n = 147)

Centre, no. (%)

Set 1 (n = 147)

Set 2 (n = 147)

≤30, no. (%)

78 (61)

91 (62)

Variables

1

48 (33)

–

>30, no. (%)

50 (49)

55 (38)

2

99 (77)

–

Median (minimum; maximum)

30 (10; 75)

30 (10; 150)

No

57 (39)

36 (25)

Yes

90 (61)

111 (75)

3

–

147 (100)

Sex, no. (%) Women

62 (42)

81 (55)

Men

85 (58)

66 (45)

Duodenopancreatectomy

145 (99)

122 (83)

Distal pancreatectomy

2 (1)

17 (12)

Surgery, no. (%)

Total pancreatectomy

–

8 (5)

Resection margins, no. (%) R0

114 (78)

110 (75)

R1

33 (22)

37 (25)

Well differentiated

64 (44)

73 (50)

Moderately differentiated

61 (41)

48 (33)

Poorly differentiated

17 (12)

22 (15)

5 (3)

4 (3)

28 (19)

24 (16)

119 (81)

123 (84)

Negative, N0

30 (20)

36 (24)

Positive, N1

117 (80)

111 (76)

Tumour size (mm)

n = 128

n = 146

Tumour stage, no. (%) T1/T2 T3

Adjuvant treatment regimens, no. (%) Gemcitabinebased chemotherapy

28 (19)

40 (27)

Gemcitabine monotherapy followed by RCT

48 (33)

38 (26)

14 (9)

33 (22)

Other adjuvant treatment RCT, Radiochemotherapy.

Differentiation, no. (%)

Unknown

Adjuvant treatment, no. (%)

Lymph nodes, no. (%)

(hENT1), one of the major mediators of gemcitabine uptake into human cells, as a predictive biomarker of gemcitabine benefit in resected and advanced PDAC has been reported in several retrospective series, including one from our group.1–6 In these studies, high hENT1 mRNA or protein levels were predictive of gemcitabine adjuvant benefit. Most of these results were obtained with a non-commercially available murine monoclonal anti-hENT1 antibody (clone 10D7G2),1–5 which led to the development of a rabbit antibody (SP120). In the recently reported randomized phase III study conducted by Poplin et al.7, hENT1 tumour expression was not found to predict gemcitabine benefit in patients with treatment-naive metastatic PDAC. Similar results were reported by Ormanns et al.8 for advanced PDAC in the AIOPK0104 phase III trial. It is of note that hENT1 expression was assessed in both studies with SP120. These results call into question the predictive value of SP120. Here, we report our experience with both antibodies, because the question of hENT1 testing in PDAC © 2014 John Wiley & Sons Ltd, Histopathology, 66, 457–462.

hENT1 testing in pancreatic adenocarcinoma

remains a hot topic in oncology and for the pathological community.

Materials and methods The study included a total of 294 consecutive and unselected patients who underwent, between September 1996 and August 2009, curative intent surgery for PDAC at three university centres with expertise in the management of PDAC (Saint-Antoine, Erasme, and Beaujon). Patients did not receive any preoperative chemotherapy or chemoradiotherapy. The patients were divided into two sets: (i) set 1, n = 147 patients from the Saint-Antoine and Erasme centres; and (ii) set 2, n = 147 patients from the Beaujon centre. We used the same tissue microarrays as used in our previously published series.2 For set 2, all patients had analysable tumour spots when tissues were stained with 10D7G2, but seven had worn-out tumour spots when tissues were stained with SP120 (140 patients analysed). The staining with 10D7G2 was performed by Mackey et al. in their laboratory, as previously described.2 SP120 (Springbioscience, Pleasanton, CA, USA) was tested in three different laboratories with Leica immunohistochemistry automate (Melbourne, Australia) (set 1: Saint-Antoine and Erasme hospitals) or Ventana Medical Systems (Tucson, AZ, USA) immunohistochemistry automate (set 2: Beaujon hospital). Briefly, sections of 4 lm from the paraffin-embedded tissue samples were cut onto silane-treated Super Frost slides (CML, Nemours, France) and left to dry at 37°C overnight. The slides were deparaffinized in xylene and rehydrated in pure ethanol. Endogenous

A

High; n = 71 High; n=5

B

Moderate/Low; n = 66

Moderate/Low; n = 76 High; n=5

Moderate/Low; n = 71

hENT1 expression on Set 2 (n = 147)

Murine 10D7G2

Rabbit SP120

peroxidase was blocked with 3% hydrogen peroxide in methanol for 30 min. Before immunostaining, antigen retrieval was performed by immersing sections in citrate buffer (pH 6.0). For set 1, sections were incubated for 10 min at room temperature with SP120 (dilution 1:100). The Bond polymer Refine Detection kit (Leica, Leica Biosystems Newcastle Ltd, Newcastle upon Tyne, UK) was used as the detection system. For set 2, a first round of staining was performed with a classic indirect biotin streptavidin system (iVIEW DAB Detection Kit; Ventana Medical Systems, Tucson, AZ) and a second round with highly sensitive synthetic, non-endogenous hapten technology with the secondary antibody (Optiview; Ventana Medical Systems). With both techniques, SP120 was diluted at 1:50. The intensity of hENT1 staining was ranked from 0 to 2, as previously described (low/moderate versus high).2 Low staining was defined as a staining intensity of 0 in >95% of tumour cells, and high staining was defined as a staining intensity of 2 in >50% of tumour cells; the remaining cases were considered to have moderate staining. Set 1 was independently scored by three separate individuals (M.S., R.M., and J.B.B.) blinded to the patient outcomes. Diagnostic reproducibility was assessed with Fleiss kappa. Set 2 was scored by expert pathologists (M.S. and J.C.) blinded to the patient outcomes. A consensus was reached for discordant cases. The hENT1 staining results were correlated with overall survival (OS) following gemcitabine treatment. hENT1 high (score 2) cases were considered to be positive, and compared with pooled hENT1 moderate and hENT1 low cases. A Cox proportional

hENT1 expression on Set 1 (n = 147)

Murine 10D7G2

Rabbit SP120

459

High; n = 57 High; n = 25

Moderate/Low; n = 30

Moderate/Low; n = 90 High; n = 34

Moderate/Low; n = 49

Figure 1. Human equilibrative nucleoside transporter 1 (hENT1) expression and concordance between the two antibodies in set 1 (A) and set 2 (B). The upper rows represent staining with 10D7G2, and the lower rows staining with SP120. Cases stained with SP120 are categorized according to their staining with 10D7G2. *In set 2, hENT1 expression could be assessed in all 147 patients with 10D7G2 but in only 138 patients with SP120. This is represented by shorter boxes in (B). © 2014 John Wiley & Sons Ltd, Histopathology, 66, 457–462.

460

M Svrcek et al.

hazards regression model was used for analysis of survival and for estimating hazard ratios (HRs) with 95% confidence intervals (CIs). The appropriate institutional review board approved this translational study.

Table 2. Predictive value of human equilibrative nucleoside transporter 1 expression according to the type of antibody in patients treated with adjuvant gemcitabine

No.

Relative hazard of mortality

Low

40

1

High

42

0.48

IHC failure

23

Antibody

Results Patient characteristics and gemcitabine chemotherapy regimens are summarized in Table 1. In set 1 (147 patients), hENT1 expression with SP120 was high in 10 cases (7%) and moderate/low in 137 cases [33 cases (22%) with moderate expression, and 104 cases (71%) with low expression]. With 10D7G2, 71 cases (48%) showed high hENT1 expression, 54 cases (37%) moderate expression, and 22 cases (15%) low expression.2 Among the 71 cases that were hENT1 high with 10D7G2, 50 were low with SP120, 16 were moderate, and five were high (Figure 1A). Interestingly, five cases that were hENT1 low with 10D7G2 were high with SP120. The interobserver agreement was excellent (kappa score: 0.89). In patients who received adjuvant gemcitabine (n = 105), there was no difference in OS between the hENT1 high and hENT1 low/moderate subgroups (HR 0.92; 95% CI 0.16–5.34; P = 0.79) with SP120, whereas high hENT1 expression was significantly predictive of OS with 10D7G2 (HR 0.48; 95% CI 0.27– 0.84; P = 0.01) (Table 2). In the second set, in which the technique without amplification was used, the results were similar, with a very low number of hENT1 high cases with SP120 (n = 15, 11%). Use of the highly sensitive polymerbased amplification technique resulted in 59 (42%) tumours being hENT1 high with SP120 [57 (38%) with 10D7G2]. However, the concordance between the two antibodies for hENT1 high was present in