Search

PDF file - 197K, Supplemental Table 1. Characteristics of GIST patients for the immunohistochemistry study. Supplemental Table 2. Characteristics of GIST and STS patients studied in flow cytometry. Supplemental Table 3: Antibodies for flow cytometry and IHC. Supplemental Figure 1. Correlation between T and NK cell infiltrates and the 3-digit Miettinen score. Supplemental Figure 2. GIST TIls are mainly composed of CD4+ T cells.

International audience; Simple Summary As patients with rare cancers face specific problems, reference networks have been developed in several European countries and then at the European level to improve their management. In some cases, the specialized centers belonging to reference networks provide remote services (specialized diagnosis review, discussion in the Multidisciplinary Tumour Board, etc.) to increase access to these services. Using data from the national sarcoma reference network implemented in France (NETSARC+), the IGeAS research program assesses the potential of its organization to address the geographical inequalities in cancer management. We analyze the individual, clinical, and geographical determinants of the overall survival of sarcoma patients in France. We found no association between the overall survival of sarcoma patients and variables measuring their social deprivation, remoteness from reference centers, and geographical context. Following previous results from the research program, this study suggests that reference network organization should be considered to reduce cancer inequalities. The national reference network NETSARC+ provides remote access to specialized diagnosis and the Multidisciplinary Tumour Board (MTB) to improve the management and survival of sarcoma patients in France. The IGeAS research program aims to assess the potential of this innovative organization to address geographical inequalities in cancer management. Using the IGeAS cohort built from the nationwide NETSARC+ database, the individual, clinical, and geographical determinants of the 3-year overall survival of sarcoma patients in France were analyzed. The survival analysis was focused on patients diagnosed in 2013 (n = 2281) to ensure sufficient hindsight to collect patient follow-up. Our study included patients with bone (16.8%), soft-tissue (69%), and visceral (14.2%) sarcomas, with a median age of 61.8 years. The overall survival was not associated with geographical variables after adjustment for individual and clinical factors. The lower survival in precarious population districts [HR 1.23, 95% CI 1.02 to 1.48] in comparison to wealthy metropolitan areas (HR = 1) found in univariable analysis was due to the worst clinical presentation at diagnosis of patients. The place of residence had no impact on sarcoma patients' survival, in the context of the national organization driven by the reference network. Following previous findings, this suggests the ability of this organization to go through geographical barriers usually impeding the optimal management of cancer patients.

387 Background: Several options have been evaluated as 2nd line treatment in mGA after failure of 1st line platinum based chemotherapy including taxanes and IRI as monotherapies or paclitaxel combined with ramucirumab but with a limited efficacy. REG monotherapy showed promising efficacy as 2nd or 3rd line treatment in mGA (Pavlakis N et al, J Clin Oncol. 2016). Methods: This comparative phase 2 multicenter randomized study was designed to evaluate the safety and efficacy of REG + IRI (REGIRI) vs IRI alone as 2nd line treatment in MGA patients (pts). Key eligibility criteria were histologically proven mGA (gastric or gastroesophageal junction (GEJ) tumor Siewert II and III), fluoropyrimidine and platinum-based first-line chemotherapy and ECOG PS ≤1. Pts received IRI 180 mg/m2 IV on D1 and D15. In REGIRI arm REG 160 mg/day was added on D2-D8 and D16-D22 every 28 days. Primary endpoint was overall survival (OS). A total of 122 events (154 pts) were required based on an assumption of 4-month (mo) gain in median OS (from 6 to 10 mo, HR=0.60), with a preplanned interim analysis of safety after 38 pts and efficacy after 40 OS events. Genotyping of cyclin-D1 was performed to evaluate the impact of treatment efficacy. Results: A total of 89 patients (REGIRI, n=44; IRI, n=45) with median age of 62 (range: 28-82) years and a mGA (67.4% with GEJ, 22.8% HER2+) were included out of the 154 initially planned by 21 sites in France. Study was prematurely stopped after interim analysis for unfavourable efficacy/toxicity ratio. With a median follow-up of 19.4 mo, 79 OS events were observed and median OS was 6.3 mo (95%CI[5.2-7.1]) with REGIRI vs 8.2 mo (95%CI[5.2-9.7]) with IRI (HR=1.11 CI95%[0.70-1.74], p=0.66). Median PFS was 2.2 mo vs 1.9 mo (NS) with REGIRI and IRI, respectively. Objective response rate was 15.9% and 13.3% respectively. Disease control rate was 45.5% and 33.3%. AEs grade ≥ 3 related to treatment were reported in 52.3% of pts in REGIRI arm vs 23.3% in IRI arm with 4 toxic deaths (diarrhea, sepsis and thromboembolic events) vs 1 (primary tumour perforation), respectively. The main severe toxicities were diarrhea: 18.2% vs 7% and febrile neutropenia: 8.2% vs 0%. Genotyping of cyclin D1 was performed on 72 pts: 19 presented A/A genotype, 38 A/G and 15 G/G. There were no differences in terms of OS according to genotyping. Conclusions: PRODIGE58/REGIRI trial was prematurely stopped after the interim analysis due to high toxicities reported in REGIRI arm with negative results on its primary endpoint. Further ancillary analysis are expected to identify toxicity risk factors and pts subgroups benefit with this combination. Clinical trial information: NCT03722108 .

Background A surveillance program was performed in colorectal cancer (CRC) patients after surgery, to diagnose asymptomatic recurrence. Aims To assess whether 18-FDG positron emission tomography/CT (PET/CT) improved the detection of recurrence during a 3-year follow-up. Methods A multicentre, two-arm randomised prospective trial comparing different 36-month follow-up strategies. Complete colonoscopy was performed at baseline and after 3 years and clinical exams with imaging every 3 months. The conventional arm (A) received carcinoembryonic antigen, liver echography, and alternated between lung radiography and computed tomography (CT) scans. The experimental arm (B) received PET/CT. Results A total of 365 patients with colon (79.4%) or rectal cancer (20.6%), stages II (48.2%) or III (50.8%), were enroled in this study. At 36 months, intention-to-treat analysis revealed recurrence in 31 (17.2%) patients in arm A and 47 (25.4%) in arm B (p = 0.063). At 3 years, 7 of 31 relapses (22.5%) in arm A were surgically treated with curative intent, compared to 17 of 47 (36.2%) in arm B (p = 0.25). The rates of recurrence and new cancers were higher in arm B than arm A (p = 0.038). Conclusions PET/CT follow-up every 6 months did not increase the rate of recurrence at 3 years or the rate of surgically treated recurrence compared with conventional follow-up.

Oral therapies have highly modified cancer patient management and changed hospital practises. We introduce a specific Oral Therapy Centre and retrospectively review information prospectively recorded by co-ordination nurses (CNs) (the DICTO programme). We describe the roles played by CNs in the management of oral cancer therapies at Limoges Dupuytren Hospital between May 2015 and June 2018. All cancers, irrespective of stage or whether oral general chemotherapy or targeted therapy was prescribed, are included. We followed up 287 patients of median age 67 years (range 26–89 years). Of these, 76% had metastases and 44% were on first-line therapy. The vast majority (88%) of their first CN contacts occurred just after physician consultation and lasted an average of 60 min. As part of follow-up, the CNs made 2719 calls (average 10 min) to patients to educate them and to verify compliance and drug tolerance. They also received 833 calls from patients (70%) or their relatives or health professionals (30%) seeking advice on management of side effects. In addition to the initial appointments, 1069 non-scheduled follow-up visits were made to assess side effects (49.2%). The CNs devoted 5 h to each patient over 3 months of treatment (i.e. 25 min/day) and, also organised scheduled hospitalisations in the department of oncology for 51% of patients. We show the interest and real-life work in a specific oral therapy centre within oncology department with the role of CNs to facilitate the global health care of the patients. Electronic supplementary material The online version of this article (10.1007/s12032-020-01393-7) contains supplementary material, which is available to authorized users.

Chemotherapy (ct) is the preferred treatment option in metastatic colorectal cancer (mCRC). The objective of the study was to determine the overall survival (OS), disease-free survival (DFS) and ct-free survival (CFS) of pulmonary thermal ablation (TA) and its place in the treatment of mCRC. All consecutive patients treated (over 11 years) with percutaneous TA for lung metastasis of colorectal origin were reviewed. All sequences of treatments were considered. We determined the OS, DFS and CFS of pulmonary TA. Two hundred and nine patients underwent 323 TA procedures for 630 lung metastases. Majority of the metastases (71.5%) were unilateral with a median diameter of 10 mm (2–46). A single metastasis was treated in 95 patients (45.5%), and 2–8 in 114 patients (54.5%). One hundred and thirty-two patients (63.2%) had only a single procedure, 77 patients (36.8%) had 2–5 procedures. Following the first TA (n = 209), 125 patients (59.8%) resumed ct. Sixty-four out of the 126 patients presenting lung progression were treated again with TA. The median CFS was 12.2 months (95% CI: 10.3–17.7). Patients with no extra-pulmonary metastases showed a statistically better CFS than those who had extra-pulmonary metastases with a median of 20.9 and 9.2 months, respectively (p

15 Background: Compared to observation, maintenance therapy with a fluoropyrimidine +/- bevacizumab showed significant improvement in progression-free survival (PFS) but not in overall survival (OS) in patients with unresectable metastatic colorectal cancer (mCRC) and disease control after first-line doublet chemotherapy (CT) +/- bevacizumab. Few studies are available on the role of maintenance therapy after induction anti-EGFR-based CT, and the benefit from anti-EGFR maintenance monotherapy during CT-free intervals (CFI) in patients with RAS wild-type (wt) mCRC. Methods: RAS wt unresectable mCRC patients with controlled disease after FOLFIRI + cetuximab (8 cycles) were randomized (1:1) to receive maintenance with bi-weekly cetuximab alone (arm A) or observation (arm B) until disease progression (PD)/unacceptable toxicity/death. Randomization was stratified according to tumor response, center, baseline Köhne Score, CEA and platelet count. In case of tumor progression during the CFI, FOLFIRI + cetuximab was to be reintroduced for 8 cycles, followed by a new CFI. Tumor response was assessed per RECIST1.1 every 8 weeks. The primary objective of this multicenter non-comparative randomized phase II trial was 6-month PFS rate after initiation of maintenance therapy. A total of 134 randomized and evaluable patients (67 per arm) were required (Fleming’s one-step design, one-sided α=5%, β=20%, H0: 40%; H1: 55%). Secondary endpoints were overall response rate (ORR), time to strategy failure, PFS, OS, safety, quality of life, circulating tumor cells and circulating tumor DNA detection and dynamic changes during treatment. Results: From January 2014 to April 2019, 214 patients were included and 139 randomized (67 arm A/72 arm B) in 35 centers. Baseline characteristics were: males, 67%/69%; median age, 64/68 years; ECOG PS 0, 54%/46%; previous adjuvant therapy, 25%/14%; single metastatic site, 58%/47%; right-sided primary, 24%/18%. The ORR in the overall and the randomized population was 55% and 72%, respectively. The median follow-up was 30 months. The 6-month PFS rate after initiation of maintenance therapy was 30% 95%CI[19; 42] in the maintenance arm, and 6% 95%CI[2;14] in the observation arm, with a median PFS of 5.3 95%CI[3.7;6.5] and 2.0 95%CI[1.8;2.8] months, respectively. Any grade treatment-related toxicity, including skin rash (40%/4%), diarrhea (33%/8%), and hypomagnesemia (46%/10%) was more frequent in arm A. Conclusions: Based on the study hypothesis, the cetuximab maintenance arm did not meet the primary objective. However, the clinically meaningful difference in PFS without any overlap in the confidence intervals between the two arms warrants further investigation. Clinical trial information: NCT02404935.

11506 Background: U-LMS and ST-LMS are rare tumors with poor prognosis when locally advanced or metastatic, and with moderate chemosensitivity. Overall response rates (ORR) given in the 1st-line setting do not exceed 50% for U-LMS and 35% for ST-LMS with a mean response duration of 3- 6 months without impact on overall survival (OS). In 2015 we reported very encouraging results of the LMS-02 study (NCT02131480) with ORR of 59.6% in U-LMS, and 39.3% in ST-LMS with manageable toxicity (Pautier; Lancet oncol 2015). Herein, we report the updated results of progression-free survival (PFS) and final results of overall survival (OS). Methods: Patients (pts) received 60 mg/m² intravenous Doxo followed by trabectedin 1.1 mg/m2 as a 3-hour infusion on Day 1 and pegfilgrastim on Day 2, repeated every 3 weeks for up to 6 cycles. Surgery for residual disease was permitted. Patients were stratified into U-LMS and ST-LMS groups. Results: Overall, 108 patients with LMS with a median age of 59 years and mostly metastatic disease (85%) were enrolled. Of those, 77 patients (71.3%) have received all 6 cycles of treatment, and 20 patients (18.5%) had metastasis resection. With a median follow-up of 7.2 years (95% CI: 6.9 - 8.2), the overall median PFS was 10.1 months (95% CI: 8.5 - 12.6), being 8.3 months (95 CI: 7.4 - 10.3) and 12.9 months (95% CI: 9.2 - 14.1) in U and ST group, respectively. Median OS was 34.4 months (95% CI: 26.9 - 42.7), being 27.5 months (95% CI: 17.9 - 38.2) in U-LMS and 38.7 months (95% CI: 31.0 - 52.9) in ST-LMS group. The median OS among the 20 pts with surgery was not reached vs 31.6 months in the population without surgery (95% IC: 23.9 - 35.4). Conclusions: The Doxo +Trab combination is an effective 1st-line therapy for pts with LMS, with promising PFS and OS results and an acceptable safety profile. Merely for comparison, the most recent results of Doxo alone in metastatic LMS, given in 1st-line setting in a phase III ANNOUNCE trial conducted during the same period, reported median PFS of 6.9 months, and median OS of 21.9 months (ASCO 2019 LBA3). Results of the LMS04 trial (NCT02997358), a randomized phase III study comparing this combination vs Doxo alone in 1st-line therapy in metastatic LMS are pending. Clinical trial information: NCT02131480 .

Natural Killer (NK) cells control metastatic dissemination of murine tumors and are an important prognostic factor in several human malignancies. However, tumor cells hijack many of the NK cell functional features compromising their tumoricidal activity. Here, we show a deleterious role of the TNFα/TNFR2/BIRC3/TRAF1 signaling cascade in NK cells from the tumor microenvironment (TME). TNFα induces BIRC3/cIAP2 transcripts and reduces NKp46/NCR1 transcription and surface expression on NK cells, promoting metastases dissemination in mice and poor prognosis in GIST patients. NKp30 engagement, by promoting the release of TNFα, also contributes to BIRC3 upregulation, and more so in patients expressing predominantly NKp30C isoforms. These findings reveal that in the absence of IL-12 or a Th1-geared TME, TNFα can be considered as a negative regulatory cytokine for innate effectors.

Objective: The study objectives were to analyze the impact of the number of lymph nodes (LNs) reported as resected (NLNr) and the number of LNs invaded (NLNi) on the prognosis of esophageal cancer (EC) after neoadjuvant chemoradiotherapy. Background: Pathological LN status is a major disease prognostic factor and marker of surgical quality. The impact of neoadjuvant chemoradiation (nCRT) on LN status remains poorly studied in EC. Methods: Post hoc analysis from a phase III randomized controlled trial comparing nCRT and surgery (group nCRT) to surgery alone (group S) in stage I and II EC (NCT00047112). Only patients who underwent surgical resection were considered (n = 170). Results: nCRT resulted in tumoral downstaging (pT0, 40.7% vs 1.1%, P 3 vs 0) 3.5, 95% CI: 2.0-6.2, P = 15) 0.95, 95% CI: 0.6-1.4, P = 0.807 and HR ( = 23) 1.4, 95% CI: 0.9-2.0, P = 0.131] did not. In Poisson regression analysis, nCRT was an independent predictive variable for reduced NLNr [exp(coefficient) 0.80, 95% CI: 0.66-0.96, P = 0.018]. Conclusions: nCRT is not only responsible for disease downstaging but also predicts fewer LNs being identified after surgical resection for EC. This has implications for the current quality criteria for surgical resection

Introduction A comprehensive geriatric assessment (CGA) evaluating several domains of health is recommended for elderly patients with cancer. Effects of altered domains on the risk of death in this population need to be clarified. The aim of this study was to estimate the independent association of each CGA domain to overall survival (OS). Method Patients included in the ONCODAGE cohort completed a CGA at baseline. Cox models (one per domain) estimated the hazard ratio (HR) of death for each CGA domain. Directed Acyclic Graphs (DAGs) selected specific sets of adjustment factors for each model. Results The analysis included 1264 patients (mean age: 78 years, women: 70%). Median follow-up was 5.2 years, and 446 patients died. Each altered domain had a detrimental effect on survival, sometimes dependent on gender, age, education or time from inclusion. Nutritional status had a time-varying effect, with higher mortality rates if altered only within the first 3 years of follow-up. In case of altered mobility, the risk of death was higher only for the youngest patients and, in case of altered autonomy, only for the youngest women. An altered neurological state led to higher mortality rates; this effect increased with the level of education. Patients with altered psychological status or more than four comorbidities at baseline had also higher mortality rates. Conclusions Patients with an altered CGA domain have a higher risk of death than those without any alteration. The effect of some alterations is different in some subgroups or at a given time of the treatments.

Purpose Although often investigated in locally advanced esophageal cancer (EC), the impact of neoadjuvant chemoradiotherapy (NCRT) in early stages is unknown. The aim of this multicenter randomized phase III trial was to assess whether NCRT improves outcomes for patients with stage I or II EC. Methods The primary end point was overall survival. Secondary end points were disease-free survival, postoperative morbidity, in-hospital mortality, R0 resection rate, and prognostic factor identification. From June 2000 to June 2009, 195 patients in 30 centers were randomly assigned to surgery alone (group S; n = 97) or NCRT followed by surgery (group CRT; n = 98). CRT protocol was 45 Gy in 25 fractions over 5 weeks with two courses of concomitant chemotherapy composed of fluorouracil 800 mg/m2 and cisplatin 75 mg/m2. We report the long-term results of the final analysis, after a median follow-up of 93.6 months. Results Pretreatment disease was stage I in 19.0%, IIA in 53.3%, and IIB in 27.7% of patients. For group CRT compared with group S, R0 resection rate was 93.8% versus 92.1% (P = .749), with 3-year overall survival rate of 47.5% versus 53.0% (hazard ratio [HR], 0.99; 95% CI, 0.69 to 1.40; P = .94) and postoperative mortality rate of 11.1% versus 3.4% (P = .049), respectively. Because interim analysis of the primary end point revealed an improbability of demonstrating the superiority of either treatment arm (HR, 1.09; 95% CI, 0.75 to 1.59; P = .66), the trial was stopped for anticipated futility. Conclusion Compared with surgery alone, NCRT with cisplatin plus fluorouracil does not improve R0 resection rate or survival but enhances postoperative mortality in patients with stage I or II EC.

Cancer immunosurveillance relies on effector/memory tumor-infiltrating CD8+ T cells with a T-helper cell 1 (TH1) profile. Evidence for a natural killer (NK) cell-based control of human malignancies is still largely missing. The KIT tyrosine kinase inhibitor imatinib mesylate markedly prolongs the survival of patients with gastrointestinal stromal tumors (GIST) by direct effects on tumor cells as well as by indirect immunostimulatory effects on T and NK cells. Here, we investigated the prognostic value of tumor-infiltrating lymphocytes (TIL) expressing CD3, Foxp3, or NKp46 (NCR1) in a cohort of patients with localized GIST. We found that CD3+ TIL were highly activated in GIST and were especially enriched in areas of the tumor that conserve class I MHC expression despite imatinib mesylate treatment. High densities of CD3+ TIL predicted progression-free survival (PFS) in multivariate analyses. Moreover, GIST were infiltrated by a homogeneous subset of cytokine-secreting CD56bright (NCAM1) NK cells that accumulated in tumor foci after imatinib mesylate treatment. The density of the NK infiltrate independently predicted PFS and added prognostic information to the Miettinen score, as well as to the KIT mutational status. NK and T lymphocytes preferentially distributed to distinct areas of tumor sections and probably contributed independently to GIST immunosurveillance. These findings encourage the prospective validation of immune biomarkers for optimal risk stratification of patients with GIST. Cancer Res; 73(12); 3499–510. ©2013 AACR.

Purpose Elderly patients form a heterogeneous population. Evaluation of geriatric factors may help evaluate a patient's health status to better adapt treatment. Patients and Methods Elderly patients with previously untreated metastatic colorectal cancer (mCRC) were randomly assigned to receive fluorouracil (FU) -based chemotherapy either alone or in combination with irinotecan (IRI) in the Fédération Francophone de Cancérologie Digestive (FFCD) 2001-02 study. Sites participating in the geriatric substudy completed geriatric screening tools to perform prognostic factor analyses for treatment safety during the first 4 months after treatment initiation. Results The geriatric score was calculated in 123 patients (44%). Median age was 80 years (range, 75 to 91 years). The Charlson comorbidity index was ≤ 1 in 75%, Mini-Mental State Examination (MMSE) score was ≤ 27/30 in 31%, and Instrumental Activities of Daily Living (IADL) showed impairment in 34% of the patients. Seventy-one patients (58%) had grade 3 to 4 toxicity, 41 (33%) had a dose-intensity reduction of more than 33%, and 54 (44%) had at least one unexpected hospitalization during the first 4 months after starting treatment. In multivariate analysis, significant predictive factors for grade 3-4 toxicity were IRI arm (odds ratio [OR], 5.03), MMSE ≤ 27/30 (OR, 3.84), and impaired IADL (OR, 4.67); for dose-intensity reduction of > 33%, the significant predictive factors were alkaline phosphates > 2 × upper limit of normal (OR, 4.16) and IRI arm (OR, 6.85); and for unexpected hospitalization, significant predictive factors were MMSE ≤ 27/30 (OR, 4.56) and Geriatric Depression Scale ≤ 2 (OR, 5.52). Conclusion Geriatric factors (MMSE and IADL) are predictive of severe toxicity or unexpected hospitalization (MMSE) in a randomized prospective phase III study in mCRC. These results suggest that cognitive function and autonomy impairment should be taken into account when choosing a regimen for chemotherapy.

A 45-year-old woman presented with a metastatic breast carcinoma and was treated with capecitabine, oral vinorelbine and trastuzumab combination therapy. The initial echocardiogram and the ECG were considered normal. She began treatment with 3-weekly cycles of the combination therapy. After the fourth dose of capecitabine, she presented with severe chest and arm pain, which was responsive to nitroglycerine spray. ECG at admission demonstrated tachycardia with ST-segment elevation suggesting ischemia. The clinical symptoms returned to baseline after a few hours and within 24 h the ECG showed inverted T in leads V3–V6. Cardiac ultrasonography revealed hypokinesia in the left ventricle without segmentary hypokinesia, with mildly reduced global systolic function, which normalized 1 week later. Two weeks later, she was rechallenged with capecitabine. After the fourth dose, the patient developed chest pain. ECG showed infero-apico-lateral injury, which normalized after administration of nitrates, nicorandil and verapamil and discontinuation of capecitabine. This case suggests that capecitabine can lead to the cardiotoxicity characteristic of other fluoropyrimidines. Therefore, it is important to inform patients about the risk of angina-like chest pain, to stop treatment immediately if symptoms occur, and to monitor the patient in hospital. Fluoropyrimidine rechallenge should be avoided because of the risk of ischemic event or sudden death.

Background Colorectal cancer (CRC) remains a major public concern. While conventional chemotherapeutic regimens have proved useful against advanced/metastatic diseases, progresses are to be made to effectively cure the large portion of patients not benefiting from these treatments. One direction to improve response rates is to develop chemosensitivity and resistance assays (CSRAs) efficiently assisting clinicians in treatment selection process, an already long preoccupation of oncologists and researchers. Several methods have been described to this day, none achieving yet sufficient reliability for recommended use in the clinical routine. Methods We led a pilot study on 19 metastatic CRC patients evaluating capacity of the Oncogramme, a standardized process using tumor ex vivo models, to provide chemosensitivity profiles and predict clinical outcome of patients receiving standard CRC chemotherapeutics. Oncogramme responses were categorized according to the method of percentiles to assess sensitivity, specificity and concordance. Results We report from a primary analysis a success rate of 97.4 %, a very good sensitivity (84.6 %), a below-average specificity (33.3 %), along with a global agreement of 63.6 % and a concordance between Oncogramme results and patients’ responses (Kappa coefficient) of 0.193. A supplementary analysis, focusing on CRC patients with no treatment switch over a longer time course, demonstrated improvement in specificity and concordance. Conclusions Results establish feasibility and usefulness of the Oncogramme, prelude to a larger-scale trial. Advantages and drawbacks of the procedure are discussed, as well as the place of CSRAs within the future arsenal of methods available to clinicians to individualize treatments and improve patient prognosis. Trial registration: ClinicalTrials.gov database, registration number: NCT02305368 Electronic supplementary material The online version of this article (doi:10.1186/s12967-016-0765-4) contains supplementary material, which is available to authorized users.