Background

Previous investigations in pancreatic cancer suggested a prognostic role for secreted protein acidic and rich in cysteine (SPARC) expression in the peritumoral stroma but not for cytoplasmic SPARC expression. The aim of this study was to evaluate the impact of SPARC expression in pancreatic cancer patients treated with gemcitabine compared with untreated patients.

Patients and methods

CONKO-001 was a prospective randomized phase III study investigating the role of adjuvant gemcitabine when compared with observation. Tissue samples of 160 patients were available for SPARC immunohistochemistry on tissue microarrays to evaluate its impact on patient outcome.

Results

Strong stromal SPARC expression was associated with worse disease-free survival (DFS) and overall survival (OS) in the overall study population (DFS: P = 0.005, OS: P = 0.033). Its negative prognostic impact was restricted to patients treated with gemcitabine (DFS: P = 0.007, OS: P = 0.006). High cytoplasmic SPARC expression also was associated with worse patient outcome (DFS: P = 0.041, OS: P = 0.011). Again the effect was restricted to patients treated with gemcitabine (DFS: P = 0.002, OS: P = 0.003). In multivariable analysis, SPARC expression was independently predictive of patient outcome.

Conclusions

Our data confirm the prognostic significance of SPARC expression after curatively intended resection. The negative prognostic impact was restricted to patients who received adjuvant treatment with gemcitabine, suggesting SPARC as a predictive marker for response to gemcitabine.

introduction

Pancreatic cancer still has to be considered as one of the most aggressive and devastating oncological diseases. New therapeutic options as the FOLFIRINOX [1] regimen and albumin-bound nab-paclitaxel [2] are emerging, but there is still a relevant and even raising mortality predicted for Europe in 2013 [3]. Also, relevant prognostic and predictive parameters are still lacking.

CONKO-001 was a randomized, controlled phase III trial investigating adjuvant therapy for patients with pancreatic cancer and established gemcitabine as standard adjuvant treatment [4]. Patients were randomized to receive either gemcitabine for 6 months or to observation only. The major finding was that gemcitabine delayed recurrence of disease for about 6 months and improved median overall survival (OS) and long-term survival [5, 6] Due to the fact that the trial includes a nontreated patient group compared with a chemotherapy group, it might be considered as an optimal starting point for translational research.

Only about 20% of patients present in resectable disease stages at the time of diagnosis. For these potentially curable cases, 5-year survival after adjuvant treatment is about 20%. Furthermore, translational research so far failed to define relevant predictive or prognostic biomarkers even though biomarker-based personalization of patient care is already reality for a number of solid cancers [7]. Over decades, the research efforts in pancreatic cancer focused on tumor cells and their molecular pathways, but, recently, the role of the peritumoral stroma and tumor–stroma interactions became an increasing matter of interest [8]. The dense desmoplastic stroma is now well recognized to be one hallmark of the disease and considered a potential barrier against chemotherapy and thereby as one reasons for therapeutic resistance.

In this context, secreted protein acidic and rich in cysteine (SPARC) is of particular interest. SPARC has been described as being overexpressed in pancreatic cancer when compared with normal tissue and to be associated with aggressive tumor behavior [9]. Previous investigations in curatively resected pancreatic cancer suggested a negative prognostic role for high SPARC expression in the peritumoral stroma but not for the cancer cell cytoplasmic SPARC expression [10].

Clinically even more important is the possibly predictive role of SPARC due to its interaction with the modified taxane nanoparticle albumin-bound (nab)-paclitaxel. In a phase I/II study in advanced pancreatic cancer, high SPARC expression was associated with a surprisingly long median OS in patients receiving a combination of nab-paclitaxel and gemcitabine in palliative first-line therapy [11].

The significance of SPARC as a target and potential predictive factor for the efficacy of nab-paclitaxel (in combination with gemcitabine) is stressed by the positive results of a recently published phase III trial in metastatic pancreatic cancer confirming a benefit for the combination therapy of nab-paclitaxel and gemcitabine compared with gemcitabine alone [2].

To our knowledge, there are no data from prospective clinical trials comparing the expression and prognostic impact of SPARC expression in patients who did not receive adjuvant chemotherapy compared with patients who received gemcitabine. The objective of our study was the evaluation of SPARC expression concerning its impact on survival in CONKO-001 patients treated with gemcitabine compared with nontreated patients. As peritumoral stroma and its constituents are believed to be critically involved in the delivery of cytotoxic drugs to cancer cells, the underlying hypothesis was that the prognostic impact of SPARC expression is restricted to patients receiving gemcitabine.

patients and methods

baseline data of CONKO-001

CONKO-001, a prospective randomized phase III study, investigated the role of adjuvant gemcitabine when compared with observation. Treatment with gemcitabine (1000 mg/m2 d1, 8, 15, q29) was continued for 6 months in an outpatient setting; subsequent follow-ups were at 8 weekly intervals. In this open, multicenter randomized, controlled trial with an active treatment arm (adjuvant gemcitabine) and a control arm (observation only), a total of 368 patients with completely resected pancreatic cancer (R0 and R1 resection) were recruited between July 1998 and December 2004. The local pathologist made the diagnosis of pancreatic ductal adenocarcinoma. All cases were re-evaluated by an experienced histopathologist before this study. Median disease-free survival (DFS) as the primary end point of the study was significantly improved by more than 6 months compared with observation only (13.4 versus 6.9 months, P < 0.001) [4]. OS was not significantly improved (22.1 versus 20.2 months) in the primary analysis, but in a re-evaluation after a longer follow-up time (22.8 versus 20.2 months, P = 0.01) [6]. The institutional review committee approved this study (trial registration isrctn.org Identifier: ISRCTN34802808).

SPARC immunohistochemistry

One hundred eighty-three tissue samples could be collect retrospectively, 160 samples were available for construction tissue microarrays (TMAs) and evaluation of SPARC expression. Diagnosis of ductal adenocarcinoma was confirmed before evaluation. Immunohistochemical staining for SPARC was carried out on TMAs according to standard procedures (clone 15G12; 1:100; Novocastra, Wetzlar, Germany). The antibody was tested for specific binding in vitro using western blotting.

Each case was represented by three tissue cores of 1 mm in diameter from different tumor areas. The slides were digitalized (Mirax Scan, Zeiss, Jena, Germany) and evaluated by virtual microscopy using the VMScope Silde Explorer (VMScope, Berlin, Germany) by two observers (MS, BVS) blinded to clinical outcome and treatment arm. Concerning stromal SPARC, the intensity of staining was evaluated by a four-tier scoring system (negative, weak, moderate and strong). In parallel to SPARC, H&E-stained TMAs were evaluated to ensure that the evaluation of stromal SPARC excluded areas of smooth muscle fibers of the duodenal wall. Concerning cytoplasmic SPARC expression, the percentage of tumor cells with staining of the cytoplasm was evaluated (0% = ‘0’, 1%–10% = ‘1’, 11%–50% = ‘2’, 51%–80% = ‘3’, 81%–100% = ‘4’). The staining intensity was evaluated as negative (0), weak (1), moderate (2) or strong (3). The numeric values were multiplied, resulting in an immunoreactivity score (IRS) ranging from 0 to 12.

The cutoff point for stromal SPARC expression (strong versus <strong) was chosen based on the data distribution alone to avoid considerable differences of sample sizes without prior knowledge of clinical outcome. Strong stromal SPARC was defined as strong staining intensity. As cytoplasmic SPARC was measured with a wider range by a 0–12 point system, the cutoff point for cytoplasmic SPARC expression (high versus low) was determined using an automated method as described earlier with the goal of optimization of prediction of DFS [12]. The definition for high cytoplasmic SPARC was IRS ≥3 (at least weak intensity in >50% of tumor cells or at least strong intensity in 1%–10% of tumor cells).

To reduce effects that might be caused by intratumoral heterogeneity, three different tumor areas were selected for the construction of TMAs. When different staining intensities were observed between the spots representing one case, the average was recorded and used for analysis. The reduce effects caused by intraobserver variability, TMAs were evaluated by two observers. Ambiguous cases were discussed to gain agreement.

statistical evaluation

DFS was defined as time from study entry to local or distant disease relapse, OS as time from study entry to death of any cause. The relation of SPARC expression with clinical and pathological tumor characteristics was evaluated using χ2 tests. The Kaplan–Meier method with log-rank tests was used for univariable survival analyses. The Cox model was used for multivariable analysis [13]. In general, P values were calculated two sided and considered as significant when <0.05.

results

Cytoplasmic SPARC expression was detectable in 95 cases (59%), in which 65 (41%) were negative. Figure 1 illustrates exemplary cases. Stromal SPARC intensity was strong in 93 cases (58%), moderate in 58 (36%), weak in 8 (5%) and negative in 1 (<1%) (Figure 2A). One hundred eighteen cases (74%) were defined as negative/low for cytoplasmic SPARC, 42 (26%) as high (Figure 2B). There was no association between stromal and cytoplasmic SPARC expression (Figure 2C).

Figure 1.

Examples of cases with negative, weak and strong stromal SPARC expression, respectively.

Figure 1.

Examples of cases with negative, weak and strong stromal SPARC expression, respectively.

Figure 2.

Frequency distribution of stromal (A) and cytoplasmic (B) SPARC expression. The cutoff points for survival analyses are indicated as dotted lines. There was no statistically significant association between stromal and cytoplasmic SPARC expression (C).

Figure 2.

Frequency distribution of stromal (A) and cytoplasmic (B) SPARC expression. The cutoff points for survival analyses are indicated as dotted lines. There was no statistically significant association between stromal and cytoplasmic SPARC expression (C).

There was no statistically significant association between SPARC expression and clinical and pathological tumor characteristics (Table 1). Despite a difference in regard to age at diagnosis (more patients >60 years in the gemcitabine group), both study arms were comparable in regard to clinical and pathological tumor characteristics (supplementary Table S1, available at Annals of Oncology online).

Table 1.

Clinical and pathological patient characteristics of the study group and association with SPARC expression

 Overall Stromal SPARC
 
Cytoplasmic SPARC
 
High, N = 93 (58.1%) Low, N = 67 (41.9%) High, N = 95 (59.4%) Low, N = 65 (40.6%) 
Age years 
 Median 62 63 62 64 60 
 Range 36–81 36–81 37–74 36–81 39–74 
Survival (months) 
 Median DFS (95% CI) 11.2 (9.2–13.3) 9.0 (5.4–12.5) 12.6 (9.0–16.2) 10.7 (7.6–14.0) 11.8 (9.1–14.5) 
 Median OS (95% CI) 21.5 (17.6–25.4) 19.8 (14.0–25.7) 26.6 (17.2–36.1) 20.4 (16.5–24.3) 26.2 (18.6–34.7) 
Karnofsky Performance Status Scale Score 
 Median 80 80 80 80 90 
 Range 50–100 50–100 60–100 60–100 50–100 
Treatment arm, n (%) 
 Gemcitabine 90 46 (51.1) 44 (48.9) 54 (60.0) 36 (40.0) 
 Observation 70 47 (67.1) 23 (32.9) 41 (58.6) 29 (41.4) 
Gender 
 Female 65 39 (60.0) 26 (40.0) 37 (56.9) 28 (43.1) 
 Male 95 54 (56.8) 41 (43.2) 58 (61.1) 37 (38.9) 
T stage 
 pT1–2 16 10 (62.5) 6 (37.5) 8 (50.0) 8 (50.0) 
 pT3–4 144 83 (57.6) 61 (42.2) 87 (60.4) 57 (39.6) 
Nodal status 
 pN0 36 21 (58.3) 15 (41.7) 24 (66.7) 12 (33.3) 
 pN1 124 72 (58.1) 52 (41.9) 71 (57.3) 53 (42.7) 
Grading 
 G1–2 92 51 (55.4) 41 (44.6) 54 (58.7) 38 (41.3) 
 G3 65 40 (61.5) 52 (38.5) 39 (60.0) 26 (40.0) 
 Missing     
Resection margin 
 R0 130 75 (57.7) 55 (42.4) 79 (60.8) 51 (39.2) 
 R1 30 18 (60.0) 12 (40.0) 16 (53.5) 14 (46.7) 
 Overall Stromal SPARC
 
Cytoplasmic SPARC
 
High, N = 93 (58.1%) Low, N = 67 (41.9%) High, N = 95 (59.4%) Low, N = 65 (40.6%) 
Age years 
 Median 62 63 62 64 60 
 Range 36–81 36–81 37–74 36–81 39–74 
Survival (months) 
 Median DFS (95% CI) 11.2 (9.2–13.3) 9.0 (5.4–12.5) 12.6 (9.0–16.2) 10.7 (7.6–14.0) 11.8 (9.1–14.5) 
 Median OS (95% CI) 21.5 (17.6–25.4) 19.8 (14.0–25.7) 26.6 (17.2–36.1) 20.4 (16.5–24.3) 26.2 (18.6–34.7) 
Karnofsky Performance Status Scale Score 
 Median 80 80 80 80 90 
 Range 50–100 50–100 60–100 60–100 50–100 
Treatment arm, n (%) 
 Gemcitabine 90 46 (51.1) 44 (48.9) 54 (60.0) 36 (40.0) 
 Observation 70 47 (67.1) 23 (32.9) 41 (58.6) 29 (41.4) 
Gender 
 Female 65 39 (60.0) 26 (40.0) 37 (56.9) 28 (43.1) 
 Male 95 54 (56.8) 41 (43.2) 58 (61.1) 37 (38.9) 
T stage 
 pT1–2 16 10 (62.5) 6 (37.5) 8 (50.0) 8 (50.0) 
 pT3–4 144 83 (57.6) 61 (42.2) 87 (60.4) 57 (39.6) 
Nodal status 
 pN0 36 21 (58.3) 15 (41.7) 24 (66.7) 12 (33.3) 
 pN1 124 72 (58.1) 52 (41.9) 71 (57.3) 53 (42.7) 
Grading 
 G1–2 92 51 (55.4) 41 (44.6) 54 (58.7) 38 (41.3) 
 G3 65 40 (61.5) 52 (38.5) 39 (60.0) 26 (40.0) 
 Missing     
Resection margin 
 R0 130 75 (57.7) 55 (42.4) 79 (60.8) 51 (39.2) 
 R1 30 18 (60.0) 12 (40.0) 16 (53.5) 14 (46.7) 

Figure 3 illustrates the survival analyses in dependence of SPARC expression. Strong stromal SPARC expression was associated with decreased patient DFS (strong versus not-strong staining intensity, 9.0 versus 12.6 months, P = 0.005) and OS (19.8 versus 26.6 months, P = 0.033). Stratified analysis according to treatment arm revealed that the effect was restricted to patients treated with gemcitabine. Patients who received gemcitabine had a shorter progression-free survival and OS when stromal SPARC was strong (DFS 12.1 versus 18.4 months; P = 0.007, OS 17.9 versus 30.2 months, P = 0.006). In the observation arm, there was no statistically significant association (DFS 6.6 versus 7.3 months, P = 0.767; OS 21.5 versus 18.2 months, P = 0.765). Cytoplasmic SPARC expression was also associated with worse patient outcome (high versus negative/low, DFS 7.4 versus 12.1 months, P = 0.041; OS 14.1 versus 25.6 months, P = 0.011). Again, the effect was restricted to the treatment arm. Patients that received gemcitabine had a worse progression-free survival and OS when cytoplasmic SPARC was high (DFS 8.3 versus 15.3 months, P = 0.002; OS 11.0 versus 28.8 months, P = 0.003). In the observation arm, there was no statistically significant association (DFS 5.8 versus 7.6 months, P = 0.844; OS 14.9 versus 20.8 months, P = 0.519).

Figure 3.

SPARC is a negative prognostic marker for disease-free and overall survival in patients treated with adjuvant gemcitabine, but not in patients that received no chemotherapy. Stromal and cytoplasmic expression as well as disease-free and overall survival are indicated by the different rows, the treatment arms by the different columns.

Figure 3.

SPARC is a negative prognostic marker for disease-free and overall survival in patients treated with adjuvant gemcitabine, but not in patients that received no chemotherapy. Stromal and cytoplasmic expression as well as disease-free and overall survival are indicated by the different rows, the treatment arms by the different columns.

In multivariable survival analysis including stromal SPARC expression, treatment arm, pT stage, pN stage, tumor grade and resection margin status, stromal SPARC expression retained its prognostic impact on DFS as well as treatment arm and tumor grading. In a second step, the analysis was repeated including cytoplasmic instead of stromal SPARC expression and cytoplasmic SPARC retained its prognostic impact on DFS as well (Table 2).

Table 2.

Univariable and multivariable proportional hazards regression for prediction of disease-free survival

 DFS univariable hazard ratio, 95% CI Stromal SPARC DFS multivariable hazard ratio, 95% CI Cytoplasmic SPARC DFS multivariable hazard ratio, 95% CI 
Stromal SPARC 
 Strong versus <strong 1.64 1.47  
1.16–2.33 1.02–2.14  
P = 0.005 P = 0.041  
Cytoplasmic SPARC 
 IRS 3–12 versus 0–2 1.48  1.61 
1.01–2.16  1.07–2.40 
P = 0.043  P = 0.022 
Treatment arm 
 Obs versus Gem 1.76 1.64 1.82 
1.26–2.48 1.14–2.35 1.29–2.57 
P = 0.001 P = 0.007 P = 0.001 
T stage 
 pT3–4 versus pT1–2 1.68 1.62 1.99 
0.91–3.12 0.86–3.03 1.05–3.78 
P = 0.099 P = 0.135 P = 0.036 
Nodal status 
 pN1 versus pN0 1.42 1.51 1.63 
0.94–2.16 0.99–2.31 1.06–2.52 
P = 0.098 P = 0.061 P = 0.028 
Grading 
 G3 versus G1–2 1.69 1.86 1.76 
1.2–2.38 1.30–2.64 1.24–2.50 
P = 0.003 P = 0.001 P = 0.002 
Resection margin 
 R1 versus R0 1.09 1.05 1.15 
0.71–1.67 0.68–1.63 0.74–1.80 
P = 0.70 P = 0.831 P = 0.53 
 DFS univariable hazard ratio, 95% CI Stromal SPARC DFS multivariable hazard ratio, 95% CI Cytoplasmic SPARC DFS multivariable hazard ratio, 95% CI 
Stromal SPARC 
 Strong versus <strong 1.64 1.47  
1.16–2.33 1.02–2.14  
P = 0.005 P = 0.041  
Cytoplasmic SPARC 
 IRS 3–12 versus 0–2 1.48  1.61 
1.01–2.16  1.07–2.40 
P = 0.043  P = 0.022 
Treatment arm 
 Obs versus Gem 1.76 1.64 1.82 
1.26–2.48 1.14–2.35 1.29–2.57 
P = 0.001 P = 0.007 P = 0.001 
T stage 
 pT3–4 versus pT1–2 1.68 1.62 1.99 
0.91–3.12 0.86–3.03 1.05–3.78 
P = 0.099 P = 0.135 P = 0.036 
Nodal status 
 pN1 versus pN0 1.42 1.51 1.63 
0.94–2.16 0.99–2.31 1.06–2.52 
P = 0.098 P = 0.061 P = 0.028 
Grading 
 G3 versus G1–2 1.69 1.86 1.76 
1.2–2.38 1.30–2.64 1.24–2.50 
P = 0.003 P = 0.001 P = 0.002 
Resection margin 
 R1 versus R0 1.09 1.05 1.15 
0.71–1.67 0.68–1.63 0.74–1.80 
P = 0.70 P = 0.831 P = 0.53 

discussion

In this study, we show that stromal and cytoplasmic SPARC expression are prognostic markers in pancreatic cancer patients treated with adjuvant gemcitabine after resection with curative intention, but not in patients that received no adjuvant therapy. To our best knowledge, this is the first evaluation of SPARC in a large randomized clinical trial including treated and untreated patients.

Our finding concerning stromal SPARC expression is in line with two immunohistochemical studies on 58 and 299 patients, respectively, that described stromal SPARC as a negative prognostic marker in pancreatic cancer [10, 14]. In addition, a study on 104 patients described SPARC mRNA expression to be associated with worse patient outcome [15]. The prognostic effect of cytoplasmic SPARC expression has not been described so far.

The strength of our study is the well-characterized study group with comprehensive follow-up data. In concordance with the ESPAC studies [16, 17], the CONKO-001 study established adjuvant gemcitabine as standard of care in pancreatic cancer [4] and contributed to underline the important role of adjuvant therapy in pancreatic cancer [18]. Due to its prospective randomized phase III status and to the fact that the study can provide data for patients treated with adjuvant gemcitabine as well as for untreated patients, the CONKO-001 study is a unique and valuable resource for translational research.

A limitation might be the fact that only tissue samples from 183 of the initial 354 CONKO-001 patients were available for this study. One hundred sixty samples were suitable for construction of TMAs, the data regarding histopathological and clinical features of the subset are comparable with the overall study population of CONKO-001. However, the sample size may be considered as a limiting factor for statistical analysis, as for example pT and pN stages were not significantly related to DFS in multivariate analysis including stromal SPARC expression indicating a possible lack of power in the study.

Another strength of our study is the straightforward and easy-to-apply four-tier scoring system based on stromal SPARC staining intensity, as simple scoring systems might reduce the interobserver variability of immunohistochemistry.

SPARC belongs to the family of matricellular proteins that are associated with the structural components of the extracellular matrix. It is a collagen-binding protein that is involved in a variety of different functions as regulation of extracellular matrix composition, regulation of matrix metalloproteinases and modulation of growth factor signaling by interaction with soluble growth factors and cellular receptors [19]. The role of SPARC in cancer is complex and appears to be tissue and context specific. Experimental data in pancreatic cancer are conflicting [20]. SPARC derived from pancreatic stellate cells has been shown to increases the invasion of pancreatic cancer cells in vitro [14]. On the other hand, in mouse models of pancreatic cancer, SPARC −/− mice present with more aggressive tumors with altered extracellular collagen deposition, reduced apoptosis, reduced blood vessel density and an increase of immunosuppressive macrophages when compared with SPARC +/+ mice [21, 22]. SPARC has also been shown to be frequently lost in pancreatic cancer cells by hypermethylation [23, 24]. The treatment of pancreatic cancer cells with secreted SPARC derived from fibroblasts has been shown to have tumor-suppressive functions [23].

One of the most interesting clinical aspects of SPARC is its potential role as a predictive marker for response to treatment with nab-paclitaxel. nab-Paclitaxel is a nanoparticle-sized albumine-bound taxane, initially developed to reduce taxane-associated toxicities. It is approved for second-line therapy in advanced breast cancer [25] and very recently for metastatic pancreatic cancer in combination with gemcitabine.

In preclinical mouse models, nab-paclitaxel leads to increased antitumor activity and higher intratumoral taxane-concentrations when compared with conventional taxane formulations [26]. In a mouse model of pancreatic cancer, the combination of nab-paclitaxel with gemcitabine increases the intratumoral concentration of gemcitabine by inhibition of gemcitabine metabolism by cytidine deaminase [27].

In patients treated with nab-paclitaxel for head and neck squamous cell cancer, the response was higher in SPARC positive tumors [28]. In breast cancer treated with nab-paclitaxel, the predictive value of SPARC is currently evaluated prospectively in the GeparSepto neoadjuvant trial (NCT01583426).

Von Hoff et al. investigated the significance of stromal SPARC expression in the context of a dose-finding phase I/II study including 37 patients for the combination of first-line nab-paclitaxel with gemcitabine. In this palliative trial, patients with a high SPARC expression treated by the combination therapy had a better outcome when stromal SPARC was high with an impressing median OS of about 18 months compared with 8 months in patients with a low SPARC expression. An additional experimental study could demonstrate higher gemcitabine levels in tumors pretreated with nab-paclitaxel and a depletion of the desmoplastic stroma. The improvement of clinical outcome may therefore not only be the effect of direct anticancer cell mechanisms of nab-paclitaxel, but additionally of targeting SPARC in the peritumoral stroma [11].

Current data of a randomized phase III trial confirmed the efficacy of nab-paclitaxel in combination with gemcitabine in the palliative first-line therapy (von Hoff, NEJM 2013). Data on SPARC expression are not available so far.

Our study underlines the important role of SPARC expression in pancreatic cancer. As a new finding, the study demonstrates that the prognostic impact of SPARC was restricted to patients who received adjuvant treatment with gemcitabine. For the patients in the observation group—that can provide data for the natural course of disease after resection and before recurrence—no difference was found for DFS or OS. The patients with strong stromal or cytoplasmic SPARC expression in the gemcitabine group had a similar poor OS compared with the patients of the observation group. These findings imply that stromal SPARC might impede the uptake of gemcitabine in pancreatic cancer and patients with high SPARC expression seem not to benefit from adjuvant gemcitabine in terms of an improved OS. This could be explained by altered diffusion of cytotoxic drugs in differentially composed tumor stroma [29].

Treatment with nab-paclitaxel might be an interesting approach as especially patients with high SPARC expression might benefit who seem to have a decreased response to gemcitabine. The predictive role of SPARC for response to nab-paclitaxel and gemcitabine should further be evaluated in clinical trials.

funding

There was no need of grant support to complete the here-presented analysis and manuscript. The CONKO-001 study was supported in part by a grant from Lilly Deutschland, Bad Homburg, Germany. CONKO-001 was an investigator-initiated trial, Lilly Deutschland had no part in the design and conduct of the trial or in the collection, analysis and interpretation of the data.

disclosure

The authors have declared no conflicts of interest.

acknowledgements

We thank Sylwia Handzik for the construction of the tissue microarrays and Kerstin Petri and Kerstin Witkowski for their excellent technical assistance.

references

1
Conroy
T
Desseigne
F
Ychou
M
et al.  
FOLFIRINOX versus gemcitabine for metastatic pancreatic cancer
N Engl J Med
 
2011
364
1817
1825
2
Von Hoff
DD
Ervin
T
Arena
FP
et al.  
Increased survival in pancreatic cancer with nab-paclitaxel plus gemcitabine
N Engl J Med
 
2013
369
1691
1703
3
Malvezzi
M
Bertuccio
P
Levi
F
et al.  
European cancer mortality predictions for the year 2013
Ann Oncol
 
2013
24
792
800
4
Oettle
H
Post
S
Neuhaus
P
et al.  
Adjuvant chemotherapy with gemcitabine vs observation in patients undergoing curative-intent resection of pancreatic cancer: a randomized controlled trial
JAMA
 
2007
297
267
277
5
Sinn
M
Sinn
BV
Striefler
JK
et al.  
Long-term survival in patients with pancreatic cancer: relevant factors in CONKO-001
J Clin Oncol
 
2012
30
4020
6
Oettle
H
Neuhaus
P
Hochhaus
A
et al.  
Adjuvant chemotherapy with gemcitabine and long-term outcomes among patients with resected pancreatic cancer: the CONKO-001 randomized trial
JAMA
 
2013
310
1473
1481
7
Crane
CH
Iacobuzio-Donahue
CA
Keys to personalized care in pancreatic oncology
J Clin Oncol
 
2012
30
4049
4050
8
Mahadevan
D
Von Hoff
DD
Tumor-stroma interactions in pancreatic ductal adenocarcinoma
Mol Cancer Ther
 
2007
6
1186
1197
9
Guweidhi
A
Kleeff
J
Adwan
H
et al.  
Osteonectin influences growth and invasion of pancreatic cancer cells
Ann Surg
 
2005
242
224
234
10
Infante
JR
Matsubayashi
H
Sato
N
et al.  
Peritumoral fibroblast SPARC expression and patient outcome with resectable pancreatic adenocarcinoma
J Clin Oncol
 
2007
25
319
325
11
Von Hoff
DD
Ramanathan
RK
Borad
MJ
et al.  
Gemcitabine plus nab-paclitaxel is an active regimen in patients with advanced pancreatic cancer: a phase I/II trial
J Clin Oncol
 
2011
29
4548
4554
12
Budczies
J
Klauschen
F
Sinn
BV
et al.  
Cutoff Finder: a comprehensive and straightforward Web application enabling rapid biomarker cutoff optimization
PLoS One
 
2012
7
e51862
13
Cox
DR
Regression models and life-tables
J R Stat Soc Ser B (Methodol)
 
1972
34
187
220
14
Mantoni
TS
Schendel
RR
Rodel
F
et al.  
Stromal SPARC expression and patient survival after chemoradiation for non-resectable pancreatic adenocarcinoma
Cancer Biol Ther
 
2008
7
1806
1815
15
Miyoshi
K
Sato
N
Ohuchida
K
et al.  
SPARC mRNA expression as a prognostic marker for pancreatic adenocarcinoma patients
Anticancer Res
 
2010
30
867
871
16
Neoptolemos
JP
Stocken
DD
Friess
H
et al.  
A randomized trial of chemoradiotherapy and chemotherapy after resection of pancreatic cancer
N Engl J Med
 
2004
350
1200
1210
17
Neoptolemos
JP
Stocken
DD
Bassi
C
et al.  
Adjuvant chemotherapy with fluorouracil plus folinic acid vs gemcitabine following pancreatic cancer resection: a randomized controlled trial
JAMA
 
2010
304
1073
1081
18
Liao
WC
Chien
KL
Lin
YL
et al.  
Adjuvant treatments for resected pancreatic adenocarcinoma: a systematic review and network meta-analysis
Lancet Oncol
 
2013
14
1095
1103
19
Bradshaw
AD
Diverse biological functions of the SPARC family of proteins
Int J Biochem Cell Biol
 
2012
44
480
488
20
Neuzillet
C
Tijeras-Raballand
A
Cros
J
et al.  
Stromal expression of SPARC in pancreatic adenocarcinoma
Cancer Metastasis Rev
 
2013
32
147
167
21
Puolakkainen
PA
Brekken
RA
Muneer
S
et al.  
Enhanced growth of pancreatic tumors in SPARC-null mice is associated with decreased deposition of extracellular matrix and reduced tumor cell apoptosis
Mol Cancer Res
 
2004
2
215
224
22
Arnold
SA
Rivera
LB
Miller
AF
et al.  
Lack of host SPARC enhances vascular function and tumor spread in an orthotopic murine model of pancreatic carcinoma
Dis Model Mech
 
2010
3
57
72
23
Sato
N
Fukushima
N
Maehara
N
et al.  
SPARC/osteonectin is a frequent target for aberrant methylation in pancreatic adenocarcinoma and a mediator of tumor-stromal interactions
Oncogene
 
2003
22
5021
5030
24
Gao
J
Song
J
Huang
H
et al.  
Methylation of the SPARC gene promoter and its clinical implication in pancreatic cancer
J Exp Clin Cancer Res
 
2010
29
28
25
Gradishar
WJ
Tjulandin
S
Davidson
N
et al.  
Phase III trial of nanoparticle albumin-bound paclitaxel compared with polyethylated castor oil-based paclitaxel in women with breast cancer
J Clin Oncol
 
2005
23
7794
7803
26
Desai
N
Trieu
V
Yao
Z
et al.  
Increased antitumor activity, intratumor paclitaxel concentrations, and endothelial cell transport of cremophor-free, albumin-bound paclitaxel, ABI-007, compared with cremophor-based paclitaxel
Clin Cancer Res
 
2006
12
1317
1324
27
Frese
KK
Neesse
A
Cook
N
et al.  
nab-Paclitaxel potentiates gemcitabine activity by reducing cytidine deaminase levels in a mouse model of pancreatic cancer
Cancer Discov
 
2012
2
260
269
28
Desai
N
Trieu
V
Damascelli
B
et al.  
SPARC expression correlates with tumor response to albumin-bound paclitaxel in head and neck cancer patients
Transl Oncol
 
2009
2
59
64
29
Feig
C
Gopinathan
A
Neesse
A
et al.  
The pancreas cancer microenvironment
Clin Cancer Res
 
2012
18
4266
4276

Author notes

MS and BVS contributed equally to the publication.