Introduction

The discovery of the biological properties of platinum salts in 1965 represented a major landmark in the history of systemic anti-cancer treatment [1, 2]. Platinum-containing chemotherapy is prescribed to 10–20% of all cancer patients, and the total number of patients who have benefited from such treatment may easily run in the millions [3]. Unfortunately, the anti-tumor effect of platinum (and other chemotherapies) is often transient and therapy resistance is one of the biggest hurdles to effective cancer therapy.

Understanding and combatting chemoresistance is thus an important area of cancer research. In addition to the various tumor cell-intrinsic resistance-mechanisms that have been identified [2], it is becoming increasingly clear that interaction between tumor cells and their surrounding microenvironment plays a key role in the development of drug resistance [4, 5]. Mesenchymal stem cells (MSCs) have received much interest in this context: MSCs are multi-potent progenitor cells with the ability to home wounded tissues, where they contribute to tissue repair and regeneration. Similarly, they can home developing tumors and interfere with tumor growth and progression [6,7,8,9,10].

We previously identified a novel MSC-characteristic that may benefit tumor cells [11]: when MSCs are exposed to platinum analogs, they secrete two specific and relatively unknown fatty acids: 12-S-hydroxy-5,8,10-heptadecatrienoic acid (12-S-HHT) and 4,7,10,13-hexadecatetraenoic acid (16:4(n-3)). Both fatty acids proved extremely potent in inducing resistance to any kind of cytotoxic chemotherapy in various tumor-bearing mouse models. Non-platinum-based chemotherapies did not induce the release of these PIFAs. Since the release of these PIFAs is known to be mediated via the cyclooxygenase 1 (COX-1) and thromboxane synthase (TXAS) pathways, it was hypothesized that inhibition of these pathways could prevent PIFA release and (thus) PIFA-mediated chemoresistance. Addition of a COX-1 or TXAS inhibitor, such as indomethacin or ozagrel, respectively, indeed prevented PIFA release and enhanced cisplatin-efficacy in mice, whereas neither indomethacin nor ozagrel alone exerted any anti-tumor effect.

If translatable to cancer patients, inhibition of these pathways could provide a target to prevent PIFA release and to enhance the clinical benefit of chemotherapy by preventing resistance. Several observations support this hypothesis [11, 12]: first, MSCs are present in circulation and tumors of cancer patients, where they may be activated. Second, platinum-stimulated human and mouse MSCs secrete the same, specific PIFAs, both capable of inducing complete chemoresistance in one xenograft- and two mouse models. Third, in cancer patients, an acute rise in plasma-PIFA levels has been observed within hours after platinum-based chemotherapy. And fourth, commonly used food supplements (such as fish oil) may contain both PIFAs [13, 14]. Altogether, these observations suggest that the MSC-mediated mechanism of PIFA-induced chemotherapy-resistance may indeed be translated to the clinic.

Given the central role of platinum analogs in many chemotherapeutic regimens, and the fact that virtually all patients with advanced disease eventually develop resistance to chemotherapy, the potential impact of preventing chemoresistance by inhibiting PIFA release is significant. No clinical trials have tested the combination of platinum-based chemotherapy and COX-1 or TXAS-inhibitors. Since indomethacin and ozagrel seemed equally capable of enhancing chemotherapy efficacy in our preclinical study, indomethacin may provide a logical candidate for further clinical exploration, given that it is a well-established non-selective COX inhibitor, which recommended dosage for analgesic, anti-flogistic purposes (75–200 mg/day) results in adequate plasma concentrations to inhibit COX-1 [15, 16]. Inhibition of COX-1, however, can reduce renal blood flow and may thereby exacerbate platinum-associated nephrotoxicity [17,18,19,20]. For this reason, concomittant use of non-steroidal anti-inflammatory drugs (NSAIDs), such as indomethacin, during platinum- (and especially cisplatin-) based treatment is generally contra-indicated. A phase I study, to evaluate the safety of combining indomethacin with platinum-based treatment and to explore its efficacy and associated PIFA levels, was therefore, conducted.

Materials and methods

Study design

The PIFA-01 study was a prospective, multi-center, open-label, phase I dose-escalation trial, conducted at three hospitals in the Netherlands and one in Switzerland (NCT01719926). The study aimed to assess the safety of indomethacin combined with cisplatin- or oxaliplatin-based regimens. Secondary and exploratory endpoints included efficacy and pharmacodynamics of such treatment, as assessed by best overall response, progression free- and overall survival and treatment-induced PIFA levels, respectively. The study consisted of two arms, based on two commonly used platinum-containing regimens: patients treated with oxaliplatin-plus capecitabine (CAPOX) were enrolled in arm I, and patients treated with cisplatin plus gemcitabine, capecitabine or 5FU were enrolled in arm II. The study was approved by the Medical Ethical Committee of the University Medical Center in Utrecht and conducted in accordance with GCP guidelines and the Declaration of Helsinki’s ethical principles for medical research [21]. Written informed consent was obtained from all study participants. Patients were included from August 2013 to June 2017 and have been observed until 1 month after end of study treatment (or death) for safety analyses, and until the 1st of October, 2017 (or death) for progression free- and overall survival analyses.

Patient selection

Main inclusion criteria included age ≥ 18 years, World Health Organisation (WHO) performance status ≤ 1 and normal organ functions, defined as glomerular filtration rate (GFR) ≥ 60 ml/min, absolute neutrophil count (ANC) ≥ 1.5 × 109/l, platelets ≥ 100 × 109/l, hemoglobin (Hb) ≥ 6.0 mmol/l, partial thromboplastin time (PTT) ≤ 1,5 × upper limit of normal (ULN), bilirubin ≤ 1.5 ULN, and aspartate transaminase (AST) and alanine transaminase (ALT) ≤ 3.0 × ULN, or ≤ 5 × ULN if liver metastases were present. All patients were required to have a histologically proven and radiologically evaluable advanced carcinoma, for which palliative treatment with one of the above-described chemotherapy regimens was indicated. Prior chemotherapy was allowed, as long as patients were platinum naïve for ≥ 6 months.

Study treatment

Patients received their first chemotherapy cycle according to standard of care (Fig. 1), to maximize safety and to provide an internal control for biomarker analyses. If GFR remained ≥ 60 ml/min and no grade 4 toxicity was observed by the end of cycle one, indomethacin was added from cycle two onwards. Indomethacin was taken orally, 8 days per treatment cycle, starting 2 days before chemotherapy infusion and continued for 5 days thereafter for steady-state purpose, reaching Cmax at time of platinum administration. Proton pump inhibiters (PPI) were prescribed for all patients from day 15 of the 1st cycle onwards and continued till the end of study treatment to prevent gastric ulcers. Patients were asked to refrain from using comedication expected to increase toxicity or change indomethacin plasma levels, such as selective serotonin reuptake inhibitors (SSRIs), angiotensin-converting enzyme (ACE) inhibitors, antithrombotics, diuretics, et cetera. All concurrent medications were recorded at each interim visit (i.e., ≥ 1× per treatment cycle). The first and lowest indomethacin dose level was 25 mg three times daily (total daily dose 75 mg). This was escalated to 50 mg three times daily for the second dose level (total daily dose 150 mg), and 75 mg three times daily for the third and highest dose level (total daily dose 225 mg).

Fig. 1
figure 1

Study flowchart. Schematic overview of study design. Treatment cycle one was administered according to standard of care for oxaliplatin plus capecitabine (arm I) or cisplatin plus gemcitabine, capecitabine or 5FU (arm II). From day 15 onwards, proton pump inhibiters were prescribed to prevent gastric ulcers. On day 19, GFR and toxicity were evaluated. If GFR remained ≥ 60 ml/min and no grade 4 toxicity was observed, indomethacin was added from day 20 until day 6 of the next treatment cycle. C treatment cycle, D day, DLT dose-limiting toxicity, PPI proton pump inhibitor

Full size image

Dose-limiting toxicities (DLTs) were assessed during cycle two. DLTs were defined as (1) a decrease in GFR > 30%, (2) peptic ulcer disease grade ≥ 3, (3) bleeding grade ≥ 3 or platelet count decrease grade ≥ 4, or (4) any relevant grade ≥ 3 toxicity considered to be related to-, or exacerbated by the addition of indomethacin. Dose reductions were not allowed in the second treatment cycle as toxicity-defined DLT; in that case, patients were taken off-study and considered non-evaluable. Maximum tolerable dose (MTD) assessment was based on the second treatment cycle-safety evaluations (i.e., DLT-period).

From treatment cycle three onwards, indomethacin was discontinued in case of (1) peptic ulcer disease grade ≥ 3, (2) GFR decrease > 30%, or if (3) the treating physician deemed further use of indomethacin contra-indicated. Chemotherapy continuation or dose modification was at the physician’s discretion.

Safety assessments

All patients underwent a complete medical history, physical examination, assessment of vital signs and WHO performance status, and laboratory analysis of hematology, coagulation, clinical chemistry and urine before treatment initiation. These assessments were repeated throughout study participation (before start of every new treatment cycle) and/or as clinically indicated. Adverse events (AEs) were classified and graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) version 4.03.

Efficacy assessments

Baseline tumor measurements were performed within 28 days of treatment initiation and every 9 weeks thereafter, or more frequently at the physician’s discretion. Overall response rate was evaluated by the investigators according to Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 [22].

Biomarker analyses

To characterize biologic activity of indomethacin, patients were asked to refrain from using other non-steroidal anti-inflammatory drugs (NSAIDs), prostaglandin-synthethase inhibitors and /or omega-3/omega-6 containing products [13, 14] from 2 weeks before treatment initiation onwards. Plasma 12-S-HHT and 16:4(n-3) levels were measured as described and validated previously [13, 23]. Plasma samples were obtained before and at 1, 2 and 4 h after start of chemotherapy infusion (T = 0, T = 1, T = 2 and T = 4, respectively; duration cisplatin and oxaliplatin infusion was 4 and 2 h, respectively), on day one of treatment cycle one and two (before and after start of indomethacin-treatment, respectively). Plasma was stored at − 80 °C until further analysis. All samples were processed within 2 h after blood withdrawal.

Statistical analyses

Since this was a phase I safety and tolerability trial, analyses were descriptive. Indomethacin dosage was escalated using a standard 3 × 3 design per study arm: per dose level, three patients were enrolled. If no DLT was observed, indomethacin dose would be escalated to the next dose level/cohort. If one out of three subjects experienced a DLT, the cohort would be expanded with three additional subjects. If two or more subjects experienced a DLT, that dose would be considered to be the toxic dose for the respective chemotherapy-indomethacin combination. A DLT rate of > 17% (i.e., > 1 out of six subjects) was considered unacceptable.

All patients who received ≥ 1 dose of indomethacin during on-study chemotherapy were included in the safety analyses (primary study endpoint). Adverse events (AEs) that were or could have been related to indomethacin were summarized by counting the number ofgrade ≥ 1, grade ≥ 2, and grade ≥ 3 AEs per patient, and determining the maximum grade per patient. These numbers were compared between indomethacin dose levels using the Kruskal–Wallis test. The percentages of patients with grade ≥ 1, grade ≥ 2 and grade ≥ 3 were compared using Fisher’s exact test.

All patients who received ≥ 2 cycles of on-study chemotherapy were included in efficacy evaluations (secondary study endpoint). Best overall response was analyzed per patient. Median progression free- and overall survival were estimated using the Kaplan–Meier method, overall and for each indomethacin dose level individually. Estimated means were compared between indomethacin dose levels using the log rank test.

All patients with valid and complete biomarker data were included for biomarker analyses (exploratory study endpoint). Linear mixed-effects models were used to assess (1) whether plasma PIFA levels increase during platinum-contain chemotherapy (cycle one only, no other fixed effects), (2) whether addition of indomethacin lowers plasma PIFA levels (cycle one vs. two as fixed effect, time as fixed-effect covariate), (3) whether the indomethacin dose level as continuous (fixed) effect was associated with PIFA (time as fixed-effect covariate) and (4) whether PIFA levels are associated with chemotherapy efficacy (cycle one and two, no other fixed effects). From this last model, only the difference between the two most extreme observed response levels was reported, i.e., partial response vs. progressive disease. In none of the models, a random slope seemed required on the basis of the (restricted) log-likelihood. PIFA measurements below the limit of detection were treated as zeros.

Results

Patient characteristics and treatment details

Between September 2013 and June 2017, 13 patients were enrolled: 12 in arm I and one in arm II. Accrual took longer than expected, partly due to Ethical Committee-imposed age-restrictions (≤ 70 years) at the beginning of the study, and partly due to the fact that the CAPOX is typically administered as first-line treatment rather than in the advanced, metastatic setting that this study focused on.

Ten patients were evaluable for safety- and efficacy analyses; baseline characteristics are shown in Table 1. The median number of prior treatment regimens was 2 (range 0–4). Evaluable patients received on-study chemotherapy for a median of 4 cycles (range 2–6). Median follow-up duration was 3.2 months (range 1.6–8.5) for safety analyses, and 7.9 months (range 3.4–24.3) for progression-free and overall survival analyses. Chemotherapy regimen was adjusted in six patients: two patients had 25% dose reductions from cycle one onwards (one due to DPD deficiency, the other due to hand–foot syndrome and neutropenia later on). Three patients had dose delays from cycle three onwards (two due to bone marrow toxicity, the other due to unforeseen dental surgery). One patient experienced an oxaliplatin-related infusion reaction in cycle three and did, therefore, not receive the full, planned dose.

Table 1 Baseline characteristics
Full size table

In arm I, three patients went off-study during cycle one due to rapid disease progression (n = 2) and worsening of pre-existing gastro-intestinal complaints (n = 1), and were, therefore, not evaluable. In the remaining nine patients, no DLTs were observed. Reasons to discontinue study treatment for these patients included disease progression (n = 5), patient preference to stop despite ongoing response (n = 2; one patient decided to participate in another trial, the other decided to continue CAPOX off-study due to increasing reluctance against oral medication), worsening of pre-existent neuropathy (n = 1), and indomethacin non-compatible co-medication (carbasalate calcium because of capecitabine-related cardiovascular toxicity (n = 1)).

The patient in arm II developed a DLT (acute kidney injury grade one) after completion of the first 8 days of combined cisplatin-indomethacin treatment, and went off-study during cycle two. According to the cohort rules, five more patients could enroll. Due to high probability of additional DLTs as estimated by the local investigators, however, inclusion came to a stop and arm II was closed prematurely due to non-accrual.

Safety

All AEs that were or could have been related to the addition of indomethacin are shown in Table 2. Nausea was the most common [observed in four patients (40%)], followed by fatigue, malaise, vomiting, ALAT-, ASAT-increase and white blood cell decrease [each observed in two patients (20%)]. Every other AE was observed in one patient only (10%). Out of the 33 AEs that were observed, 24 were grade one (73%), eight were grade two (24%), and one was grade three (3%). Grade four or worse AEs were not observed. Indomethacin-relatedness was classified by local investigators as ‘unlikely’, ‘possibly’, ‘probably’ or ‘definitively’ in twelve, seventeen, one and three cases (36, 52, 3 and 9%, respectively). The number of grade ≥ 1, grade ≥ 2 or grade ≥ 3 AEs per patient did not differ between indomethacin dose levels (p = 0.09, 0.5, and 0.5, respectively), nor did the maximum grade AE per patient (p = 0.6), nor did the number of patients with grade ≥ 1, grade ≥ 2 or grade ≥ 3 AEs (p = 1, 0.7 and 1, respectively).

Table 2 Toxicity and efficacy
Full size table

Two serious AEs (SAEs) were observed, both in patients who went off-study before indomethacin initiation: one patient was hospitalized due to a liver abscess, caused by primary disease progression, which ultimately led to the patient’s death. Another patient was hospitalized due to worsening of pre-existent gastro-intestinal complaints, which resolved without sequela. To minimize the risk of recurrent gastro-intestinal complaints, however, it was decided to continue chemotherapy off-study (without addition on indomethacin). No other serious adverse events were observed and no unexpected or treatment-related deaths occurred; all deaths were caused by primary disease progression.

Efficacy

Response data are summarized in Table 2. No complete responses were observed. Three patients (30%) experienced a partial response; two at dose level one (one in each arm) and one at dose level two, arm I. Stable disease was observed in three patients (30%), one at each dose level in arm I. The remaining four patients (40%) had progressive disease at first response evaluation. Median progression free- and overall survival were 5.5 months (95% CI 0.0–11.9) and 15.1 months (95% CI 1.8–28.4), respectively; differences between indomethacin dose levels were non-significant (p = 0.4 and 0.9, respectively).

Biomarkers

Plasma PIFA levels of all biomarker-evaluable patients (n = 9) and change therein upon chemotherapy infusion and indomethacin administration, as well as association with chemotherapy efficacy, are shown in Fig. 2 and Table 3. During the first cycle of chemotherapy, before addition of indomethacin, both PIFA levels seemed to increase in the 4 h after chemotherapy infusion, but differences did not reach significance (p = 0.09 for 12-S-HHT, and p = 0.2 for 16:4(n-3)). Upon addition of indomethacin, however, 12-S-HHT levels were significantly reduced across all time points (i.e., before start of chemotherapy up to 4 h thereafter: − 0.81 nmol/l, 95% CI − 1.20 to − 0.43, p = 0.0001). Furthermore, each indomethacin dose level-increase was associated with a significant drop in 12-S-HHT level (− 0.33 nmol/l, 95% CI − 0.51 to − 0.15, p = 0.0005). For 16:4(n-3), no effect of indomethacin was observed between treatment cycles (p = 0.2), nor between indomethacin dose levels (as a continuous variable, p = 0.5).

Fig. 2
figure 2

PIFA levels on day one of treatment cycle one and two. a Plasma 12-S-hydroxy-5,8,10-heptadecatrienoic acid (12-S-HHT) levels in nmol/l per patient (y-axis), measured before and on 1, 2 and 4 h after oxaliplatin infusion (x-axis). top row indomethacin 25 mg TID, middle row 50 mg TID, lower row 75 mg TID, cycle 1 without indomethacin, cycle 2 with indomethacin. b Plasma hexadeca-4,7,10,13-tetraenoic acid (16:4(n-3)) levels in nmol/l per patient (y-axis), measured before and on 1, 2 and 4 h after oxaliplatin infusion (x-axis). top row indomethacin 25 mg TID, middle row 50 mg TID, lower row 75 mg TID, cycle 1 without indomethacin, cycle 2 with indomethacin

Full size image
Table 3 Biomarker analyses
Full size table

No association was found between PIFA levels and response in the three-category partial response (PR), stable disease (SD) and progressive disease (PD) (p = 0.3 for 12-S-HHT and p = 0.2 for 16:4(n-3) on three degrees of freedom), neither after combining PR and SD into a group with ‘clinical benefit’ (results not shown). Only the difference between the two most extreme responses observed (PR vs. PD) was reported (Table 3).

Discussion

The primary aim of this study was to evaluate the safety of combined indomethacin and oxaliplatin (arm I)- or cisplatin (arm II)-based chemotherapy. In addition, we aimed to explore the effect of indomethacin on efficacy of chemotherapy and PIFA levels, and the relation between PIFA levels and chemotherapy efficacy.

First of all, combined indomethacin and cisplatin/oxaliplatin treatment was generally well tolerated. The vast majority of AEs was mild (73% grade one, 24% grade two, 3% grade three), and ‘unlikely’ (36%) or ‘possibly’ (52%) related to indomethacin. Definitive indomethacin relatedness was rare (9%). One patient in arm II experienced a DLT, due to > 30% decrease in GFR after completion of the first 8 days of combined indomethacin-cisplatin treatment. This was in line with expectations, given the impact of NSAIDs on renal clearance of cisplatin [17,18,19,20]. Although the effect on renal function was mild (acute kidney injury grade one) and although the patient recovered without sequela, the reluctance amongst the investigators to include more patients in this arm was substantial. Therefore, we can not draw conclusions regarding the safety of combined indomethacin-cisplatin treatment. Combined indomethacin-CAPOX treatment, however, proved to be safe: no DLTs were observed in any of the enrolled patients, the MTD was not reached, and no significant difference in grade one, two or three AEs were observed between indomethacin dose levels. Thus, indomethacin 75 mg TID administered 8 days around chemotherapy infusion is safe in patients treated with CAPOX, and forms the recommended phase II dose.

Another study aim was to explore the effects of platinum-containing chemotherapy (with or without indomethacin) on plasma PIFA levels, and to explore PIFA levels in relation to response to chemotherapy. Based on our previous findings [11, 12], platinum-containing chemotherapy was expected to cause a rise in PIFA levels. Addition of indomethacin was expected to counteract this effect, thereby enhancing chemotherapy efficacy. In line with our preclinical studies, 12-S-HHT was indeed effectively blocked by indomethacin. Unexpectedly, however, 16:4(n-3) levels in patients did not reflect the preclinical findings. A potential explanation could be that 16:4(n-3), a fatty acid for which limited studies are available, is not exclusively produced via COX-1, or that it can be found in foods other than fish oil as well [24].

This prompts the question whether reduction of 12-S-HHT alone is sufficient to prevent PIFA-induced chemotherapy resistance, given that both PIFAs affected chemotherapy efficacy in mice. In the present study, however, no significant associations were found between PIFA levels and response to chemotherapy. Yet, our study was not designed (and thus underpowered) to test for differences in chemotherapy efficacy. In addition, the fact that five out of nine biomarker-evaluable patients discontinued study treatment for other reasons than progressive disease, may have limited our ability to correlate PIFA levels with chemotherapy efficacy.

Meanwhile, synergy between indomethacin- and platinum-based chemotherapy has been described before: in mice-bearing murine colorectal cancer tumors, for example, synergy was observed with cisplatin, but not with adriamycin [25]. In human lung cancer xenografts, combined indomethacin-oxaliplatin treatment increased tumor-inhibition rates in one study [26] and reduced expressions of lymph node metastasis related factors in another [27]. Further, indomethacin enhanced cisplatin efficacy in human uterine cervical and ovarian cancer cells [28, 29]. In addition, in a randomized trial including 22 pet dogs with naturally occuring invasive bladder cancer, cisplatin plus piroxicam (a COX inhibitor, similar to indomethacin) led to a remission rate of 71%, compared to 0% for cisplatin alone (p < 0.002) [30]. Altogether, these observations suggest that indomethacin can indeed enhance chemosensitivity; possibly by reduction of baseline 12-S-HHT levels alone. In addition, the fact that indomethacin is a known COX-1 inhibitor, whereas other trials to date have mainly focused on (and found little effect of) adding COX-2 inhibitors to chemotherapy [31, 32], supports the notion that indomethacin in anti-cancer treatment merits further research: present study provides evidence that it would be safe to do so. Furthermore, it proofs that patients using indomethacin for analgesic or anti-flogistic purposes can safely continue indomethacin during CAPOX treatment.

In summary, combined CAPOX- and indomethacin treatment was well-tolerated. Compared to regular CAPOX, addition of indomethacin up to 75 mg TID hardly increased overall toxicity, nor renal toxicity at any grade. Combined cisplatin-indomethacin treatment, however, did cause renal toxicity in the one patient receiving such treatment. Indomethacin significantly reduced plasma 12-S-HHT levels, but had no effect on 16:4(n-3) levels; additional research is needed to assess the effects on chemotherapy efficacy. The recommended phase II dose was established at 75 mg, taken three times daily, 8 days per treatment cycle, starting 2 days before chemotherapy infusion and continued for 5 days thereafter.