Low dose Naltrexone for induction of remission in inflammatory bowel disease patients

A prospective cohort of patients with therapy refractory IBD (CD or UC) that started LDN therapy was formed. The decision to start LDN therapy was made by the treating physician, after fully informing the patient of the possible benefits and drawbacks. All patients were prescribed 4.5 mg Naltrexone once daily. Patients were instructed to administer one dose of LDN before bedtime.

Upon initiation of LDN therapy, patients were followed according to usual care at the outpatient clinic, with contact (in person or via telephone) after 4, 8 and 12 weeks. During these visits self-assessed disease activity and adverse events were recorded. Patients were offered endoscopic evaluation and assessment of laboratory values after 12 weeks of treatment or at time of discontinuation of LDN therapy, whichever occurred earlier. This study was approved by the Ethical Committee of the Erasmus MC (MEC-2014-656).

During the follow-up, demographic data (e.g. age, gender) and IBD related data (e.g. year of diagnosis, concomitant and previous IBD related therapies, Montreal phenotype classification) were recorded. Additionally, where available, data on diagnostic tests, particularly endoscopic evaluations, performed prior to the start of LDN therapy (with a 1 week window) and during the follow-up period were recorded. All patients that completed at least 1 assessment of disease activity were included in the cohort.

Clinical outcomes were based on patient self-assessments and outpatient assessments, where available. Patients were considered non-responders if no clinical improvement occurred in the first 4 weeks of LDN therapy. Patients were considered to have clinical response if self-assessed disease activity decreased within the first 4 weeks of LDN therapy, and lasted for at least 4 weeks in total. Of secondary interest were the rates of adverse events during LDN therapy. Endoscopy results were scored based on the most severe area of inflammation. Endoscopic findings in all IBD patients were scored on a scale from 0 to 3, representing no inflammation to severe inflammation respectively.

Colorectal cancer cell lines HCT116 and CACO-2 were cultured in Dulbecco’s Modified Eagles Medium (DMEM, Lonza, Basel, Switzerland) supplemented with 100 U/mL penicillin, 100 mg/mL streptomycin (Life technologies, Bleiswijk, NL) and 10% Fecal Calf Serum (FCS, Sigma-Aldrich, St. Louis, USA). Cells were maintained at 37 °C in a 5% CO₂ humidified setting.

Non-inflamed intestinal biopsies were collected from two IBD patients undergoing routine endoscopy for their disease. Organoids were prepared as described [25, 26], see Additional file 1 for details.

Cell viability was assessed using MTT assays as described [27], see Additional file 1: Methods. Each experiment was performed twice in triplicate.

Wound healing assays were performed as described [28], see Additional file 1. The concentration of Naltrexone used was based on in vivo dosages (4.5 mg per ± 60 kg bodyweight). Experiments were performed thrice in duplicate, with two measure-sites per scratch.

Western blotting was performed as described [29], with modifications (see Additional file 1). HCT116 and CACO-2 cells were treated with Tunicamycin (2 μM), lipopolysaccharide (10 μg/mL LPS) or E. coli (paraffin-fixed DH5α, 6.25e5/mL) in the presence or absence of 1 μg/mL Naltrexone. Organoids were treated with LPS in the presence or absence of 1 μg/mL Naltrexone. Experiments were performed at least twice.

FFPE tissue sections were immunohistologically stained for GRP78, as described [17, 30], see Additional file 1. Antigen retrieval was performed by boiling the slides in 600 mL of 10 mM sodium citrate buffer, pH 6.0 for 15 min. Slides were blocked by incubating in 10% goat serum in PBS and incubated with GRP78 antibody (BiP, Cell Signaling Technology, Danvers, MA) diluted in blocking buffer (1:100) overnight at 4 °C. Rabbit envision (DAKO, Heverlee, Belgium) was used as secondary antibody.

We used rt-PCR to determine MOR expression on the IEC cell lines, using Ribosomal protein (RP2) primers were used as control [26], see Additional file 1.

Cells were plated at 0.2 × 10⁶ per well in 24 wells plates. Upon attachment to the plate, cells were treated as described in the text and supernatant was harvested after 24 h. Experiments were performed twice, in duplicate. Cytokine levels in supernatants from IECs and patient sera were determined by ELISA (Ready-SET-Go!^® eBioscience, San Diego, CA) as per manufacturer’s instructions. All samples were tested in duplicate in the ELISAs.

Continuous variables were reported as medians with interquartile range (IQR). Comparisons in continuous variables were performed with the Mann–Whitney U test. For comparisons of categorical variables, Fisher’s exact test was used. For in vitro and ex vivo experiments, normality of distribution was assessed with D’agostino and Pearson Omnibus normality test. When passing normality test or when there were insufficient numbers to calculate normality, parametric testing was performed, otherwise, non-parametric tests were employed. Student T-tests were performed for comparisons of two groups. For comparisons of more than two groups, ANOVA with post hoc testing (Tukey’s multiple comparison test) was performed. For all tests, one or two-sided (as appropriate) p-values < 0.05 were considered statistically significant, graphs show mean ± SEM. Analyses were performed using Graphpad Prism 5.0.

From July 2010 till August 2014, 47 patients were treated with LDN, of which 19 (40.4%) were male and 28 were female. Median treatment and follow-up duration after start of LDN was 3 months (IQR 3–5 months). Of the 47 patients, 28 (59.6%) were diagnosed with CD and 19 with UC. Three patients had previously undergone surgery (2 ileocecal resections and 1 subtotal colectomy, all in CD patients). The full baseline patient characteristics are described in Table 1.

All participants were either steroid dependent or steroid refractory and had previously been treated with at least one other drug, and all patients showed clinical signs of disease activity at initiation of LDN therapy. Notably, 41 patients (87.2%) had previously received at least one anti-TNFα agent, and 19 (40.4%) had been treated with two anti-TNFα agents. The 6 patients not exposed to anti-TNFα had refused anti-TNFα therapy due to fear of possible side effects. The full details on the previous and concomitant treatments at start of LDN therapy are described in Table 2. Seven patients (14.9%) reported adverse events due to LDN, including vivid dreams (N = 4), drowsiness (N = 2) and headache (N = 1). Two patients discontinued LDN therapy after 2 weeks, due to drowsiness. Vivid dream complaints improved when LDN was administered in the morning instead of at bedtime.

Of the 47 patients, 35 (74.5%) achieved a clinical response. Of those 35 patients, 12 patients had a response of at least 3 months (25.5% of total cohort, 8 CD, 4 UC), whereas a short-lived (between 4 and 12 weeks) improvement was seen in the remaining 23 patients (48.9% of total cohort, 13 CD, 10 UC). There was no statistically significant difference between CD and UC patients in the number of patients that achieved either response or remission (p = 1.000 and p = 0.515 respectively).

The median endoscopic score at baseline amongst all patients was 2 (IQR 1.25–2.0). In 12 patients, consecutive endoscopies were performed both at baseline and at 12 weeks or at time of relapse, whichever occurred earlier. These consecutive endoscopies were performed in 6 patients with response (3 CD, 3 UC) and 6 patients with remission (1 CD, 5 UC). Between these two groups, no significant difference was observed in baseline endoscopic score (median 1.67, range 1–3 versus 1.83, range 0–3 for response and remission respectively p = 0.676). However, patients achieving clinical remission had a significantly greater improvement in endoscopic score compared to patients not reaching clinical response (median change − 1.5, range − 2 to 0 versus 1.0, range 0–2, p = 0.005, Fig. 1). Complete endoscopic remission upon treatment with LDN was seen in 5 out of 6 patients with clinical remission.

Having shown clinical effect of Naltrexone in patients, we next sought to establish whether Naltrexone has a direct effect on epithelial cell function. After testing for the presence of MOR (Fig. 2a), we investigated the effect of Naltrexone on wound healing in layers of HCT116 and CACO2 colonic epithelial cell lines. Figure 2b shows that scratch wounds inflicted in HCT116 cell cultures are healed significantly faster when cells are treated with Naltrexone as compared to vehicle control (p = 0.0001 at t = 24; p = 0.0001 at t = 48). In CACO2 cells, which migrate much faster than HC116, all wounds were healed at t = 48 and the effect of Naltrexone on wound size was less clear (t = 24 h, p = 0.085, Fig. 2b, right panel). Possibly, the lower MOR expression levels observed in CACO2 cells accounts for the lesser effect of Naltrexone in this cell line. However, when comparing the migrated distance of cells, it was evident that Naltrexone stimulated epithelial cell migration, in both HCT116 (361 ± 24 vs 656 ± 52 pixels, p = 0.085) and CACO2 (465 ± 26 vs 310 ± 50 pixels, p = 0.0083, Fig. 2c). This effect of Naltrexone on wound healing was not due to an increased cellular proliferation, as total viable cell numbers were not affected by Naltrexone up to a concentration of 100 μg/mL (Additional file 2: Figure S1A–C).

In vivo, epithelial cells produce an array of cytokines in response to inflammatory stimuli, which in turn can attract immune cells and perpetuate inflammation in IBD patients. We therefore investigated whether cytokine production by epithelial cells is directly affected by Naltrexone. Cells were stimulated with bacteria in the absence or presence of Naltrexone, and supernatants were tested for the presence of the pro-inflammatory cytokines IL-6, IL-8 and TNFα after 24 h. As shown in Fig. 3, treatment of cells with bacteria significantly increased IL-8 production in both HCT116 and CACO2 cells (44 ± 4 to 92 ± 14 pg/mL in HCT116, p = 0.0001, Fig. 3a and 17 ± 1 to 25 ± 3 pg/mL in CACO2, p = 0.0001, Fig. 3b). However, neither basal levels nor bacteria-stimulated levels of IL-8 were significantly affected by Naltrexone treatment. IL-6 and TNFα levels were undetectable (not shown).

Next we tested whether IL-8 or TNFα systemic levels in patients were modulated by Naltrexone treatment in vivo. Paired serum samples (before and after initiation of treatment) were available of 7 patients, 3 responders and 4 non-responders. IL-8 was detected in 6 patients (Fig. 3c), whereas TNFα could be measured in serum from 5 out of 7 patients (Fig. 3d). No consistent up or down modulation of either IL-8 or TNFα was observed, nor were there significant differences in these cytokine levels between responders and non-responders to Naltrexone (Fig. 3e, f). Together, these data suggest that Naltrexone does not positively impact on inflammation through modulation of intestinal epithelial cell cytokine production.

As ER stress in the mucosa has been associated with the development of IBD, and Naltrexone was previously shown to reduce ER stress-induced inflammation in a model of liver damage, we next investigated whether Naltrexone has a direct effect on ER stress in intestinal epithelium. ER stress was chemically induced in intestinal epithelial cell lines by Tunicamycin (Fig. 4a and Additional file 3: Figure S2), as demonstrated by a strong upregulation of the ER stress marker GRP78. Interestingly, Naltrexone was able to reduce these levels in both cell lines. Chemical stimulation of cells with Tunicamycin causes inhibition of the UDP-N-acetylglucosamine-dolichol phosphate N-acetylglucosamine-1-phosphate transferase (GPT) and subsequent accumulation of unfolded glycoproteins in the ER; a non-physiological process likely to result in much higher ER stress levels than are probable in vivo. To investigate ER stress in a more physiologically relevant setting, we incubated intestinal epithelial cells with bacteria (representative examples shown in Fig. 4b and Additional file 3: Figure S2A). A significant upregulation of GRP78 expression was observed in HCT116 cells treated with E. coli (0.065 ± 0.007 to 0.097 ± 0.01, p = 0.0025, Fig. 4b), which again was significantly reduced by co-treatment of cells with Naltrexone (0.076 ± 0.005, p = 0.0025). In CACO2 cells, Naltrexone diminished bacteria-induced GRP78 expression in three out of three experiments, although this did not reach statistical significance as bacteria-induced ER stress was low in these cells (Additional file 3: Figure S2A). In order to further confirm ER stress pathway activation with a different physiological stimulus, we also induced ER stress in HCT116 cells with lipopolysaccharide and investigated expression levels of CHOP, a downstream target of the ER stress pathway. Figure 4c shows that CHOP levels induced by LPS were reduced by co-treatment with Naltrexone (1.315 ± 0.592 to 0.801 ± 0.710, p = 0.027). Again, the effect was less clear for CACO2 cells (Additional file 3: Figure S2C).

While cell lines are an easy and common model system to study epithelial cell function, such cell lines may show different cellular effects due to transforming mutations. We therefore generated organoids from colonic biopsies from two IBD patients, representing IBD epithelial tissue. Stimulation of these organoids with LPS again induced GRP78 expression (4 out of 4 experiments), and ER stress levels were reduced by co-treatment of organoids with Naltrexone (Fig. 4d, 1.734 ± 0.473 to 1.017 ± 0.698, p = 0.046). Together, these data imply that ER stress in intestinal epithelial cells is alleviated by Naltrexone.

Next, we investigated intestinal ER stress in patients treated with LDN. Intestinal tissue biopsies were available in 13 patients prior to treatment and in 5 patients 3 months into treatment, with 3 paired samples. Sections were stained for GRP78 (for specificity of the staining, see Additional file 4: Figure S3). High levels of ER stress were observed in both the inflamed intestinal lamina propria and crypts from IBD patients (Fig. 5a). GRP78 levels decreased upon NTX treatment, most noticeably in the lamina propria, (1.14 ± 0.5 vs 0.8 ± 0.5 for lamina propria and 0.9 ± 0.4 vs 0.7 ± 0.6 for crypts) although statistical significance was not reached because of low patient numbers (Fig. 4b). However, in the 3 paired samples available, NTX treatment reduced interstitial ER stress (Fig. 4c, and examples in Fig. 4d). In toto, these data suggest that Naltrexone has a direct effect on ER stress as measured by GRP78 expression in intestinal mucosa.