Multiplex Proximity Ligation Assay to Identify Potential Prognostic Biomarkers for Improved Survival in Locally Advanced Pancreatic Cancer Patients Treated With Stereotactic Body Radiation Therapy

Abstract

Purpose

To explore seromarker levels for associations with outcomes in locally advanced pancreatic cancer (LAPC) patients who received chemotherapy and stereotatic body radiation therapy (SBRT).

Methods and Materials

Serum from LAPC patients in 2 prospective trials of hypofractionated SBRT (5–6.6 Gy × 5) was collected before SBRT. Proximity ligation assay quantified the expression levels of 36 pancreatic cancere—specific candidate seromarkers: Axl, BMP2, CA 125, CA 19–9, CEA, CXCL-1/6/9/10, EGFR, Gas6, Her2, IGF-2, IGFBP-2/3/7, IL-6/6Ra/7/8/12, mesothelin, MMP-1/2/3/7, osteopontin, PDGFRa, PDK1, PF4, RegIV, SPARC, TGF-β, VEGF-A/D, and YKL40. Seromarker values were log transformed owing to log-normal distribution of the values, and Cox regression analysis was performed to assess for any association with overall survival. The Benjamini-Hochberg method was used to control for a false discovery rate (FDR) of only 10%.

Results

Sixty-four patients with LAPC were included. No clinical factors (including surgical resection, receipt of pre-SBRT chemotherapy, receipt of post-SBRT chemotherapy, performance status, and age) or potential biomarkers in the panel were associated with improved survival in this cohort after application of th eFD Rcorrection. Potentia lprognostic factors for improved survival for future investigation included surgical resection (P=.007, adjusted P=.153) and the serum expression of IL-8 (P=.006, adjusted P=.153), CA 19–9 (P=.031, adjusted P=.377), and MMP-1 (P=.036, adjusted P=.377).

Conclusions

These data explore the expression of a panel of proteins in pre-SBRT serum of LAPC patients in the context of a conservative FDR correction. None of the clinical factors or expression levels of the serum proteins were found to be associated with survival; however, IL-8, CA 19–9, and MMP-1 were highlighted as possible candidates warranting inclusion in future seromarker studies in the ongoing efforts to identify tools for risk stratification and treatment allocation in LAPC.

Summary

Noninvasive blood-based biomarkers can potentially stratify pancreatic cancer patients for more personalized therapies and clinical trials, because only some patients derive long-term benefit from specific therapies. We surveyed a panel of proteins to test for associations with improved overall survival in patients who received chemotherapy and hypofractionated SBRT. The results of this hypothesis-generating study can help narrow the selection of candidate seromarkers in the ongoing efforts to stratify patients according to potential response to therapy.

Introduction

Pancreatic cancer is now the third leading cause of cancer-related death in the United States (1). One-third of patients will present with unresectable locally advanced pancreatic cancer (LAPC), which has a median survival of 5 to 15 months, despite aggressive combined modality therapy (2).

Our goal in this study was to identify serum biomarkers prognostic of survival and disease progression after treatment with hypofractionated stereotactic body radiation therapy (SBRT) using proximity ligation assay (PLA), an immunopolymerase chain reaction—based method for detecting and quantifying serologic antigens that has a great dynamic detection range and has been used to develop pancreatic cancer—specific multiplexed panels (3–5).

We tested whether the expressions of 36 serum proteins in a pancreatic cancer—specific PLA panel would allow clinicians to stratify LAPC patients treated with hypofractionated SBRTaccording to expected survival at diagnosis. This study is a secondary correlative study using the baseline serum of patients treated on either of 2 clinical trials of hypofractionated, 5-fraction SBRT (NCT01146054, NCT01781728). In the first trial (NCT01146054), patients received gemcitabine followed by 33 Gy, with resumption of gemcitabine following completion ofSBRT (6). In themore recent second trial (NCT01781728), patients received systemic therapy followed by 25–33 Gy (7).

Methods and Materials

Patient selection

Histologically confirmed LAPC patients treated at Johns Hopkins Hospital (Baltimore, MD) on NCT01146054 and NCT01781728 between 2012 and 2016 were included in this study after institutional board review. Specific eligibility criteria for the trials are discussed elsewhere (6, 7).

Treatment intervention

All patients received SBRT at Johns Hopkins Hospital. Chemotherapy administration was allowed at institutions outside of study centers. Fiducial placement, radiation therapy simulation, treatment planning, and dose constraints have been previously described (6).

At the time of this study, a total of 34 and 37 LAPC patients were treated on NCT01146054 and NCT01781728, respectively. Of these, 34 and 30 patients on NCT01146054 and NCT01781728, respectively, had baseline serum collected within 6 weeks of starting SBRT and were included in this analysis (Table 1). Median follow-up of the final cohort was 12.2 months (range, 2.1–45.2 months). Stereotatic body radiation therapy was delivered in 5 daily fractions (fx) to a maximum of 33 Gy/5 fx in 48 patients (75%) and 25 Gy to 32.5 Gy/5 fxin 16patients (25%), with the latter group receiving a reduced dose to enable achievement of predefined radiation therapy dose constraints (6).

Table 1.

Patient demographics and treatment characteristics (N=64)

Patient and treatment characteristics	n (%)
ECOG performance status
0	22 (34)
1	40 (63)
2	1 (2)
Missing	1 (2)
Baseline serum sample
Before any therapy	8 (12.5)
After chemotherapy, before SBRT	56 (87.5)
Time from baseline serum to SBRT (wk), median (range)	2 (0–5.4)
Receipt of pre-SBRT chemotherapy
None	6 (9)
Any chemotherapy	58 (91)
Gem	28 (48)
EFX	17 (29)
Gem/NP	8 (14)
Other	5 (9)
No. of cycles of pre-SBRT chemotherapy, median, (range)	1 (0–10)
SBRT dose
3300 cGy/5 fx	48 (75)
<3300 cGy/5 fx	16 (25)
Post-SBRT chemotherapy
Resume within 1 mo	49 (75)
Observation	17 (25)
Surgical resection
Yes	17 (25)
No	49 (75)

Abbreviations: ECOG = Eastern Cooperative Oncology Group; FFX = fluorouracil, leucovorin, oxaliplatin, and irinotectan; fx = fractions; Gem = gemcitabine; Gem/NP = gemcitabine/nab-paclitaxel; SBRT = stereotactic body radiation therapy.

Values are number (percentage) unless otherwise noted.

Sample collection

Pre-SBRT blood samples were centrifuged and aliquoted, and plasma was stored at —80° C. Multiplex PLA, probing for 36 potential tumor markers, was performed on plasma samples. To ensure uniformity in analysis, all pathologic samples were sent to an expert at Stanford Univeristy School of Medicine to perform PLA.

Proximity ligation assay

Candidate biomarkers were identified from published data. From the Gene Expression Omnibus (3), expression levels of putative seromarkers were identified, and k-means clustering was used to identify subcategories from which potential candidates were selected (8). The entire panel used in this study is shown in Table 2 and includes 36 candidate seromarkers.

Table 2.

Pancreatic cancer—specific biomarker panel

Growth factor related	Immune modulators	Pro-metastatic	Miscellaneous
EGFR	CXCL-1	Ax1	CA 125
HER2	CXCL-6	BMP-2	CA 19–9
IGF-2	CXCL-9	Gas6	CEA
IGFBP-2	CXCL-10	MMP-1	PDK1
IGFBP-3	IL-6	MMP-2	SPARC
IGFBP-7	IL-6Ra	MMP-3	YKL40
Mesothelin	IL-7	MMP-7
PDGFRa	IL-8
RegIV	IL-12
TGF-β	Osteopontin
VEGFa	PF4
VEGFd

Abbreviations: BMP = bone morphogenic protein: CA 125 = cancer antigen 125; CA 19–9 = carbohydrate antigen 19–9; CEA = carcinoembryonic antigen; CXCL = C-X-C motif ligand; EGFR = epidermal growth factor receptor; Gas6 = growth arrest specific 6; HER2 = human epidermal growth factor receptor 2; IGF = insulin-like growth factor; IGFBP = IGF binding protein; IL = interleukin; MMP = matrix metalloproteinase; PF4 = platelet factor 4; PDGFR = platelet-derived growth factor receptor; PDK = pyruvate dehydrogenase kinase; ReglV = regenerating gene IV; SPARC = secreted protein acidic rich in cysteine; TGF = transforming growth factor; VEGF = vascular endothelial growth factor; YKL40 = chitinase-3-like protein 1.

Statistical methodology

Pre-SBRT plasma expression levels of potential biomarkers were measured using PLA and log base 2 transformed to follow a normal distribution. Overall survival (OS) was calculated as the time from end of SBRTuntil death from any cause. Patients who did not experience death were censored at the time of last follow-up. Cox regression analysis was performed to test for the association of clinical factors and biomarkers with OS. We used the Benjamini-Hochberg method for false discovery rate (FDR) adjustment to control for an FDR of <10%, chosen to reflect the exploratory nature of the study (9). In total, 41 comparisons were made for the OS endpoint: 36 seromarkers and 5 clinical factors (surgical resection,receiptof pre-SBRT chemotherapy, receipt of post-SBRT chemotherapy, performance status, and age).

Statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC), and all tests performed were 2-sided tests with an alpha level of 0.05.

Results

The median OS of the entire cohort was 13.6 months (95% confidence interval 8.0–19.2 months). Of the clinical factors evaluated, only surgical resection approached a significant association with improved OS, but it failed to meet the FDR threshold of 10% (hazard ratio [HR] 0.38, P=.007, adjusted P=.153). Of the 36 potential biomarkers evaluated, higher expression of interleukin-8 (IL-8) (HR 1.27, P=.006, adjusted P=.153), carbohydrate antigen 19–9 (CA19-9) (HR 1.22, P=.031, adjusted P=.153), and matrix metalloproteinase-1 (MMP-1) (HR 1.19, P=.036, adjusted P=.377) approached a significant association with inferior OS but also failed to meet the FDR threshold (adjusted P). These data are depicted in Table 3.

Table 3.

Cox proportional hazard model of clinical features and seromarker associations with overall survival

Clinical feature/seromarker*	Hazard ratio	P	Adjusted P
Surgical resection	0.381	.007	.153
Female sex	0.999	.0511	.396
Performance status	1.138	.664	.845
Age	0.999	.961	.996
IL-8	1.271	.006	.153
CA 19–9	1.220	.031	.377
MMP-1	1.192	.036	.377
IL-7	1.550	.057	.396
PDK1	0.715	.078	.412

Abbreviations as in Tables 1 and 2.

Only seromarker analyses with P≤.100 are shown.

^*Pre-SBRT plasma expression levels of potential biomarkers were measured using proximity ligation assay and log base 2 transformed to follow a normal distribution.

Discussion

We explored the prognostic potential of a number of biomarkers in LAPC patients. Previous reports of hypofractionated SBRT in LAPC demonstrated variable efficacy, with survival ranging from <6 months to >2 years (6, 7). To identify patients most likely to benefit from therapy, we conducted PLA analysis on pretreatment serum of LAPC patients treated with chemotherapy and SBRT on 2 separate clinical trials. We highlight a few interesting potential biomarkers, including IL-8, CA 19–9, and MMP-1, for incorporation into larger future studies.

After application of the FDR correction, successful surgical resection was not associated with improved survival in this cohort of patients. This is likely due to the limitation of our small sample size in the study, because surgical resection of LAPC has consistently been shown to be associated with improved survival (10). Thus, candidate biomarkers that similarly approached significance in this study warrant further consideration in larger cohorts.

To date, several studies have explored these potential biomarkers with clinical outcomes in pancreatic cancer patients. The most commonly describedis CA 19–9, with higher values portending inferior OS (11, 12). Interleukin-8 and MMP-1 have also been described as potentially prognostic in pancreatic cancer. Expression of C-X-C motif chemokine receptor 1, a receptor for IL8, has been associated with cancer stem-cell properties of pancreatic cancer cells and correlated with lymph node metastasis rates and poor survival (13). Excess IL-8 secretion by cancer-associated fibroblasts has been linked to pancreatic cancer invasion and metastasis (14). Similarly, MMP-1 expression has been associated with poor prognosis (15, 16). These data further suggest the continued exploration of associations of serum expression of these potential biomarkers with clinical outcomes in future studies.

Challenges and limitations of this study are rooted in its exploratory nature. In particular, the sample size is small, and there is variability in type and duration of chemotherapy received. Determining the appropriate statistical tools to account for these challenges is a common difficulty in biomarker discovery and is particularly acute in less prevalent diseases such as pancreatic cancer. Given these considerations and the number of seromarkers tested, the Benjamini-Hochberg method was used to control for an FDR of <10% (9).

In conclusion, we conducted a hypothesis-generating survey of a panel of potential biomarkers in a cohort of LAPC patients. The results of this study could serve as a foundation for future larger studies to guide the selection of biomarkers for further investigation, with the eventual goal of delivering precision medicine in this challenging disease.