1. Introduction
Pancreatic ductal adenocarcinoma (PDAC) is often a lethal condition and is ranked as the seventh highest cause of cancer related mortality in the world [
1]. Of the newly diagnosed patients, 80–85% have locally advanced or metastatic disease [
2]. Metastatic disease is characterized by a poor prognosis with a 5-year survival of less than 5% [
2] and palliative chemotherapy is the only treatment option for this patient category [
2].
The combination of 5-fluorouracil, leucovorin, irinotecan, and oxaliplatin (FOLFIRINOX) was the first major step forward in palliative systemic treatment since the introduction of gemcitabine monotherapy. Overall survival (OS) and quality of life (QOL) were significantly improved compared to gemcitabine monotherapy (11.5 vs. 6.8 months and QOL improvement of 10 points) [
2,
3,
4,
5]. Gemcitabine combined with nab-paclitaxel is one of the other currently used chemotherapy regimens. Survival was significantly improved with this combination compared to gemcitabine monotherapy (8.5 versus 6.7 months), while grade 3–4 toxicity was not increased [
2,
6,
7]. Nowadays, FOLFIRINOX and gemcitabine with or without nab-paclitaxel are widely used first-line regimes, but there is limited evidence for second-line treatment for metastatic PDAC, especially after FOLFIRINOX [
8,
9].
One of the reasons for treatment resistance of PDAC is the desmoplastic reaction initiating the production of large amounts of tumor stroma [
10,
11,
12,
13]. Stroma limits the vascularization of tumor cells, which restricts the effective delivery of anti-cancer agents to the tumor [
14]. The Hedgehog signaling pathway is known to be involved in tumor stroma formation in PDAC [
14,
15]. PDAC cells produce an increased amount of the Sonic Hedgehog ligand (SHh) [
14,
15,
16,
17,
18]. By stimulating the patched 1 receptor, the ligand initiates the desmoplastic reaction, resulting in activation of the Hedgehog signaling pathway transcription factors Gli1,2,3 by Smoothened (SMO). Elevated production of the SHh ligand results in large amounts of tumor stroma and restricts vascularization [
15,
16,
17,
18]. A study in gemcitabine-resistant mouse models showed that by targeting the Hedgehog pathway, tumor vascularization increased, initiating higher efficacy of the chemotherapeutic treatment. Indeed, when combining gemcitabine with Hedgehog inhibition, tumor vasculature and subsequently gemcitabine delivery in the tumors were enhanced [
13,
19,
20,
21]. LDE225 is a pharmacological Hedgehog signaling pathway inhibitor and is thought to reduce the amount of tumor stroma.
Over the years, trials on Hedgehog inhibition (e.g., IPI-926, vismodegib) in combination with gemcitabine monotherapy, gemcitabine+nab-paclitaxel, or FOLFIRINOX showed no statistical difference in drug delivery or treatment efficacy [
22,
23]. However, there has not yet been a trial evaluating the effect of LDE225 in combination with gemcitabine+nab-paclitaxel.
In order to establish early signs of efficacy, we incorporated two tumor microenvironment imaging techniques in our study: Intravoxel-incoherent motion modelled diffusion-weighted magnetic resonance imaging (IVIM-DWI MRI) and dynamic contrast-enhanced (DCE) MRI. IVIM-DWI can non-invasively assess tumor diffusion and perfusion in vivo [
24,
25]. Low diffusion is typically associated with dense cell structures as in solid tumor and stroma whereas increased diffusion is associated with necrosis [
26]. DCE MRI further probes the tumor’s micro vascularity and vascular permeability [
27]. Our hypothesis is that a reduction in stroma caused by LDE225 leads to increased diffusion [
28,
29] and to revascularization of the tumor showing increased perfusion [
30,
31] and can be evaluated using these quantitative imaging techniques. Additionally, necrosis of the tumor as an overall result of the treatment is expected to show an increase in diffusion [
28]. In a previous work, we already optimized IVIM-DWI and DCE MRI specifically for PDAC patients [
32,
33]. Using optimized pipelines, we correlated both IVIM-DWI and DCE MRI to pancreatic cancer pathology and treatment response in PDAC patients receiving surgery and illustrated the response in IVIM-DWI in patients receiving chemo-radiotherapy [
26,
34]. This highlights the potential of these techniques for evaluating treatment in PDAC patients.
Our current study is the first to explore the modification of the desmoplastic reaction seen in pancreatic cancer using two approaches, targeting tumor stroma by nab-paclitaxel and the Hedgehog inhibitor LDE225 and targeting the tumor cells with gemcitabine and nab-paclitaxel as a second-line treatment for patients with metastatic PDAC after first-line FOLFIRINOX.
2. Patients and Methods
2.1. Patient Population
Patients registered in this study were 18 years of age or older with histologically or cytologically confirmed diagnosis of metastatic PDAC and provided written informed consent. All patients had measurable disease on a pre-treatment CT scan according to response evaluation criteria in solid tumors (RECIST) 1.1, a World Health Organization (WHO, Geneva, Switzerland) performance status <2, and adequate bone marrow and organ function. Patients were excluded if they had a history of hypersensitivity to LDE225, or to drugs of similar chemical classes. Additionally, patients who underwent previous treatment with smoothened inhibitors or with known central nervous system (CNS) metastases were excluded.
2.2. Study Design and Treatment
This was a multi-center, open-label, interventional, noncontrolled, nonrandomized dose finding, phase I/II study, conducted in the Amsterdam University Medical Centers, location AMC in Amsterdam and in the Medical Spectrum Twente hospital in Enschede, both in the Netherlands. The study was approved by the ethical committee and registered at ClinicalTrials.gov with the identifying number NCT02358161. This study was conducted in agreement with the latest revision of the declaration of Helsinki and with the guidelines of good clinical practice issued by the European Union.
The objective of the phase I part of the study was to assess the safety, maximum tolerated dose (MTD), and dose limiting toxicities (DLTs) of LDE225 when co-administered with fixed doses of gemcitabine and nab-paclitaxel. The objective of the phase II part was to evaluate the efficacy and safety of the treatment combination after prior FOLFIRINOX treatment with response rates according to RECIST 1.1, median overall survival (OS) and progression free survival (PFS), changes in vascularity with DCE MRI, and changes in tumor stroma with DWI MRI.
At the start of our study, the largest study published on second-line treatment with gemcitabine and nab paclitaxel in patients with metastatic pancreatic cancer was the study by Portal et al. [
35]. In this study, an objective response rate of 17% was seen. For the sample size calculation of the phase II part of the trial, we hypothesized that if the combination could lead to a response rate of 20%, developing a randomized trial is reasonable. With a power of 80% to detect such an increase and a significance level (alpha) of 0.10, the minimum sample size needed was 27 evaluable patients. Anticipating 10% of patients not being available for analysis, we planned to include a total number of 30 patients.
The starting dose of LDE225 was 400 mg daily dosed orally [
36]. The doses for nab-paclitaxel and gemcitabine were 125 mg/m
2 and 1000 mg/m
2, respectively, administered weekly for three weeks every 4 weeks [
6].
A DLT was defined as any dose limiting toxicity that was considered related to LDE225 alone or in combination with nab-paclitaxel and gemcitabine and unrelated to disease progression, inter-current illness, or concomitant medications (see
Table 1). A minimum of three patients were entered on each dose level and followed for six weeks. Subsequent enrolment of new cohorts was based on the toxicity assessment in the first cycle and the documentation of any DLTs (see
Table 2). If 0 out of 3 patients experienced a DLT at a given dose level, 3 patients were entered at the next dose level (+200 mg LDE225). When 1 out of 3 patients experienced a DLT, 3 patients were entered at the same dose level. Dose escalation was stopped when more than 2 patients experienced a DLT at a certain dose level. This dose level was declared the MTD.
2.3. Toxicity Assessment
Toxicity was graded using the common terminology criteria for adverse events (CTCAE) version 4.0. Before every treatment with gemcitabine and nab-paclitaxel, adverse events (AEs) were scored and reported in the case report file (CRF).
2.4. Tumor Response Evaluation
At baseline and subsequently every 8 weeks, tumor assessment and evaluation according to RECIST 1.1 was performed using CT-scan. (non)Target lesions were measured per organ side and documented in the CRF.
Overall response as well as response to (non)target lesions were described as complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD). If there were any new lesions compared to earlier screening, this was also documented in the case report file and regarded as PD.
IVIM-DWI and DCE MRI were performed at baseline (prior to treatment) and 8 weeks after the start of chemotherapy on a 3T MRI scanner (Ingenia, Philips, Best, The Netherlands). For IVIM-DWI MRI, a diffusion-weighted multi-slice echo-planar imaging sequence was used with TR/TE 2200/45 ms, respiratory triggering, and 12 b-values from 0 to 600 [
33,
34]. To minimize bowel movement, 20 mg of hyoscine bromide (Buscopan, Boehringer, Ingelheim, Germany) were administered intravenously before the acquisition. Detailed relevant MRI sequence parameters for all scans are given in
Table S1 in the Supplementary Materials. The IVIM model was fitted to the signal decay of the DWI MRI as a function of the b-values using a bi-exponential fit to obtain the diffusion (D), pseudo diffusion (D*), and perfusion fraction (f) maps.
DCE MRI was performed identically to our previous work [
32]. We acquired a dynamic series of 3-D spoiled gradient echo images with temporal resolution of 1.75 s, TR/TE 3.2/2.0 ms, and FA 20°. Scans were repeated for 280 s and after 10 dynamics, and 0.1 mmol/kg of 1.0 mmol/mL gadobutrol (Gadovist, Bayer Healthcare, Leverkusen, Germany) was injected intravenously at 5 mL/s followed by a 15 mL saline flush. Prior to the DCE acquisition, a Look-Locker ultrafast gradient echo was performed to assess the baseline T1 values, which were used to determine the contrast concentration. A population-based arterial input fraction was derived from another dataset of pancreatic cancer patients using the same scan settings and contrast administration protocol [
32,
37]. The Tofts model was fitted voxel-wise to acquire the extracellular extravascular space (EES) volume fraction (v
e), the fractional plasma volume (v
p), the transfer rate of contrast from plasma to EES (K
trans), and the reflux rate of contrast from EES to plasma (k
ep) [
38].
The primary tumor was manually delineated on the baseline and post-treatment MRI scans using a 3D Slicer (Available online:
http://www.slicer.org, accessed on 7 September 2021) under guidance of a contrast-enhanced MRI from the same scan session and a contrast-enhanced CT scan [
39]. Cancerous pancreatic tissue was included into the region of interest (ROI) and biliary stents were excluded from the ROI. The mean parameter values of DCE and IVIM-DWI MRI from within the ROI were used for further analysis.
MRI data of patients from the phase I and phase II part of the trial were all combined to analyze the influence of LDE225 combined with gemcitabine and nab-paclitaxel on the characteristics of the tumor. A total of 36 patients underwent a baseline MRI scan of which 23 patients also underwent a post-treatment MRI scan (see
Table S1).
2.5. Statistical Analysis
Data in this study were analyzed using IBM SPSS software version 22. Baseline characteristics were described using mean (standard deviation) or median (interquartile range) for continuous variables and absolute number (percentage) for categorical variables. Evaluation of adverse events, safety, and efficacy of LDE225 combined with gemcitabine and nab-paclitaxel was performed with descriptive statistics. A Kaplan Meier analysis described the median OS between the different dose levels and treatment groups. Data analysis was anonymous. The probability of a type-I error was set at 0.05.
All statistical tests in the response evaluation using DCE and IVIM-DWI MRI were two-tailed and a significance level of α = 0.05 was used. The overall effect of the chemotherapy on the tumor was assessed by a Wilcoxon signed-rank test between MRI scans at baseline and post-treatment for all DCE and IVIM-DWI parameters. Subsequently, a receiver operating characteristics (ROC) analysis was performed to determine the specificity and sensitivity (using the Youden’s index) of baseline MRI parameters and the relative change in parameter value during treatment to predicting OS of PDAC patients receiving chemotherapy. The mean OS of 222 days was taken as a cut-off value to divide the patient group in long and short OS for the purpose of the ROC analysis.
The baseline and post-treatment CA 19.9 levels in combination with the relative change in MRI parameter values during treatment were also used to evaluate the treatment response. This was assessed with the spearman’s rank correlation coefficient.
4. Discussion
In this phase I/II trial, which mainly focused on FOLFIRINOX-pretreated patients with metastatic pancreatic cancer, LDE225 in combination with gemcitabine and nab-paclitaxel demonstrated a manageable safety profile and promising efficacy. The overall response rate (ORR) and durability of response compares favorably with outcomes provided with currently available therapy for this population.
The reason for this focus was that in the patient cohort that was previously treated with FOLFIRINOX in the phase I part of the trial, despite the lower dose of LDE225, we saw responses, which was unprecedented at that time. Therefore, we decided to focus on the post-FOLRIRINOX group in the phase II part of the trial. Indeed, the evidence for second-line treatment after failure on FOLFIRINOX is scarce. Since FOLFIRINOX is the recommended first-line treatment for patients with metastatic PDAC in The Netherlands, there is a desperate need for a suitable second-line treatment after FOLFIRINOX.
There are a few randomized clinical trials in advanced pancreatic cancer, but they all have been conducted after first-line gemcitabine-based chemotherapy. The most promising combination in this setting is liposomal irinotecan in combination with 5-FU/LV, demonstrating a median survival of 6.1 months versus 4.2 months for the 5-FU/LV single agent [
40]. After failure on FOLFIRINOX, data on second-line treatment are sparse. Although a gemcitabine-based regimen combined with nab-paclitaxel might be an option, randomized trials to confirm this suggestion are lacking. In the ACCORD/ PRODIGE 4 trial, about 50% of patients underwent second-line treatment with gemcitabine, with a median OS of 4.4 months, which is less favorable compared to an OS of 6 months in our clinical trial [
4]. Other studies describing treatment with gemcitabine and nab-paclitaxel after FOLFIRINOX failure found lower median OS compared to our study [
41,
42,
43]. Currently, there is no randomized evidence available on second-line treatment with gemcitabine and nab-paclitaxel after FOLFIRINOX failure. Observational cohort studies on second-line treatment with gemcitabine and nab-paclitaxel after FOLRIRINOX treatment in first line showed ORR of 13% and 17% [
35,
44,
45]. These might be comparable to our ORR of 13%, but as opposed to other phase II/III studies on metastatic PDAC patients, median PFS and ORR in our study were higher [
40,
46,
47,
48]. The combination treatment of LDE225 with gemcitabine and nab-paclitaxel showed an improved biologic activity and was safely tolerated. However, the non-controlled design does not permit any conclusions, and future phase III clinical trials should confirm these results.
In The Netherlands, FOLFIRINOX is currently the recommended first-line treatment for patients with metastatic PDAC [
8,
49]. For patients who are not eligible for FOFIRINOX in first line, it would be interesting to preselect patients for LDE225 in combination with gemcitabine and nab-paclitaxel by using MRI (lower baseline perfusion fraction results in higher OS) in future studies.
The adverse events observed in our study were different from phase I studies with LDE225 monotherapy in patients with advanced solid tumors of any kind, including medulloblastoma and basal cell carcinoma. These studies most commonly found fatigue (2.3%), anorexia (2.3%), and elevated creatine phosphokinase (CPK) levels (4.7%) [
36,
50,
51]. The difference in adverse events might be attributable to the addition of gemcitabine and/or nab-paclitaxel. However, although the incidence of adverse events is higher compared to previous studies, the toxicity was manageable enough for patients to continue treatment.
We were able to detect treatment effects from combined LDE225, gemcitabine, and nab-paclitaxel using quantitative MRI. We showed that the fractional plasma volume and diffusion of the tumor increased during treatment. Two mechanisms might contribute to this increase: apoptosis as a result of the chemotherapy reaching the tumor and the decrease of stroma due to LDE225. The lower cellularity due to these two processes causes a higher diffusion [
28,
52]. Various studies also described an increase of diffusion in tumors due to chemotherapy [
53,
54]. We excluded one patient from the analysis because of necrosis at baseline resulting in outliers in baseline D values. In this specific case, the response of the tumor to chemotherapy is expected to be different, and the necrotic cells will be cleared and less tumor cells will become necrotic.
Furthermore, we found that the baseline perfusion fraction can be used to predict OS. In patients with lower baseline perfusion fraction, the OS was higher. Additionally, an increase in perfusion fraction during treatment resulted in a better prognosis. These results can be explained by the treatment with LDE225, which specifically targets the tumor stroma. Patients with tumors that have a higher level of stroma at baseline will show a lower baseline perfusion fraction. The relative reduction of stroma by LDE225 will be higher in these patients than in patients with a lower amount of stroma at baseline. Our findings highlight the importance of assessing the tumor microenvironment with DCE and particularly IVIM-DWI during treatment. Furthermore, we showed that these techniques may allow for precision medicine by selecting patients most likely to benefit from LDE225.
A limitation of this study is that all patients who received MRI scans, also patients with LDE225 doses other than 200 mg, were included in the MRI analyses. Since all other analyses (e.g., on OS and PFS) were only performed on patients receiving 200 mg LDE225, there could be some discrepancy between these results. In addition, the studied patients is a very selected group, because metastatic PDAC with a WHO performance status of 0 or 1 after pre-treatment with FOLFIRINOX is remarkable.