The design of this trial was three-armed, randomized, double blinded and placebo controlled, consisting of 14 days of treatment and 1 month of follow-up, totaling to a length of 45 days. It was conceptualized as a pilot study to gain knowledge about dose and efficacy of NBMI in mercury-intoxicated individuals. Furthermore, the feasibility of a future phase-IIb study at the site was examined. The three arms consisted of two treatment groups, treated with 100 mg/d NBMI and 300 mg/d NBMI, respectively, and one placebo group. The subjects were randomized into the treatment arms in equal numbers to obtain a 1:1:1 ratio.
The study subjects were recruited from the local gold miner population, since the artisanal miners in the region of Zaruma/Portovelo, Ecuador, were occupationally exposed to mercury by handling it directly to amalgamate the gold dust or smelting amalgam to extract the gold. The miners were approached in collaboration with local mining organizations and through advertising in the local community. An informed consent to participate in the study was obtained from all participants. Urine samples of 865 interested miners were collected and analyzed for their mercury content. The purpose of this prescreening was to identify individuals with mercury in urine values > 15 μg/l for further screening. Of 90 persons with mercury urine levels exceeding the threshold 44 were examined whether they met all the inclusion criteria and none of the exclusion criteria until the sample size of 36 was reached (see Appendix and Fig. 2). For inclusion in the study, subjects had to fulfil the following criteria: an age between 18 and 65 years, urine-Hg ≥ 15 μg/l, a mercury intoxication medical score of ≥ 5 or ≥ 3 in combination with at least two of the additional symptoms social nervousness/withdrawal, irritability, memory loss, metallic taste, mental- and physical fatigue.
All the participants obtained an emergency card with information on the investigational product, the subject’s id, the investigator’s name, an emergency number as well as the name and address of the sponsor. The whole study was conducted at the Hospital Doctor Humberto Molina, Zaruma, Ecuador, where direct access to medical emergency care could be provided.
There were no formal sample size calculations for this study, since the main goal was to explore the feasibility for a more extensive study. The size of 36 subjects was chosen as a generally accepted number for a pilot study. 12, 11 and 13 subjects in the treatment groups meant that the 95% confidence intervals for the difference between mean values in a NMBI treated group and the placebo treated group resulted in a length of about 1.6 times the standard deviation.
The three treatment arms differ in participant numbers because after randomizing the first 19 individuals, the necessity to relabel some of the boxes with an extended expiry date occurred which caused an interruption of the process. The subjects corresponded to three blocks of six and the remaining subject was allocated as an incomplete block. With the arrival of the relabeled batch, the randomization was continued with new blocks and caused the inequality in group size. However, this way the subjects could be randomized and keeping the blind of the study.
The participants were analyzed according to the treatment arm they were assigned to, regardless of the treatment received. The analysis included all participants who took at least one dose of medication (NBMI or placebo) and at least one post baseline assessment of the MIS. This analysis set corresponds to a modified intention-to-treat (mITT) analysis set. The per protocol (PP) analysis set includes only participants without major protocol violations. This set was used as a sensitivity analysis for the efficacy endpoints.
As shown in Fig. 2, 36 subjects were selected for the study. 34 (94.4%) participants were administered the assigned daily dose of the study product on site for all 14 days of treatment. Two subjects took only one capsule for the first 2 and 3 days, respectively, based on an administrative error in the hospital. The following 12 and 11 days all doses were taken as planned by both subjects. One of those two subjects had been allocated to the placebo arm and missed no active dose.
For the 14 days of the intervention 100 mg NBMI, 300 mg NBMI or placebo were administered orally to all patients. The medication comprised of three capsules per administration for each group, two containing 50 mg and one containing 200 mg, either of the investigational product, placebo or a combination of both in the following regime:
300 dose arm: two 50 mg NBMI capsules and one 200 mg NBMI capsule,
100 mg dose arm: two 50 mg NBMI capsules and one 200 mg placebo capsule,
Placebo dose arm: two 50 mg placebo capsules and one 200 mg placebo capsule.
The use of placebo filled dummy capsules ensured the blinding of the patients. No restriction on medication intake in relation to meals was set. Other medication necessary for the patients’ well-being was allowed to be administered at the investigator’s discretion, but recorded appropriately. The subjects were asked to abstain from drinking energy drinks containing taurine or glucuronolactone and alcohol equivalent to more than two half-liter bottles per day 72 h prior to the medical assessment visits at day 1, 6 ± 1, 15 and 45 ± 1. Moreover, for the length of treatment, the subjects were to refrain from the intake of herbal medicine, grapefruit or grapefruit juice to avoid possible drug interactions. As the use of mercury in mining is banned in Ecuador no restriction to working in the mines were made.
Randomization and assignment to treatment groups
Upon enrolment, each participant was assigned a consecutive number. The subjects were allocated to the treatment arms (100 mg/d, 300 mg/d or placebo) at a ratio of 1:1:1. For this purpose, a randomization list was created with the subject numbers using the R system and the subjects were allocated to the three arms in blocks of six.
At the start of the treatment, the participants were handed out numbered boxes with the capsules corresponding to their subject number. Each participant received the same number of blisters, either filled with NBMI, placebo capsules or both. The boxes and the blisters containing the investigational product and/or placebo were labelled with a subject number by an external packaging company to ensure the blinding of the investigators.
The treatment compliance was controlled and documented by the use of investigational product accountability logs. In those logs the date and quantities of study products received, dispensed to and used by each subject and products returned and destroyed at the end was recorded. All administration of the study drug was supervised, and each participant received the daily dose at the clinic. The recorded data was verified by the study monitor during the study’s process and at the end of the follow up, all investigational products used or returned were accounted for.
The main aim of this study was to generate explorative and descriptive data as there is no consistent method for efficacy assessment or guideline for treatment of mercury intoxication. No formal primary efficacy endpoint was predetermined, and the outcomes were specified as differences in comparison to baseline values and the placebo treated subjects. As described in previous reports, the following characteristics were examined: the mercury intoxication medical score (MIS) and its individual components . Moreover, a panel of neuro-motoric functions (CATSYS) , mental and physical fatigue score , tremor (finger-to-nose test)  and mercury values in plasma and urine were analyzed.
The samples for analyzing the mercury concentration in plasma and urine were taken at day 1 (pre-dose), 15 and 45 and sent to the qualified laboratory ALS Scandinavia AB in Luleå, Sweden for analyzing.
Medical intoxication score
The medical intoxication score (MIS) is a tool created for identifying mercury intoxication in patients. It is a ten-point score which is assessed through medical examination, neuromotoric tests and an anamnestic questionnaire [3, 5, 6]. The items consist of excessive salivation, tremor at work, sleeping problems, bluish discoloration of gingiva, ataxia, dysdiadochokinesia as much as a heel-to-shin test, a match box test, a pencil tapping test and additionally a dipstick test for Proteinuria. Each one can have an assigned value of 0 or 1, depending on whether the symptom is absent (0) or present (1) or whether the test result is negative (0) or positive (1). The mercury intoxication medical score is the sum of the items’ values.
In this study, a change in the score and the value of its individual items compared to baseline assessment was used as an efficacy parameter. Accordingly, the change value for the score ranges from − 10 (maximal improvement of symptoms) to 7 (maximal worsening of symptoms), since the least score for subject inclusion is 3. The change value for the individual items can be − 1 (for worsened condition), 0 (for unchanged condition) and 1 (for improved condition). The mercury intoxication medical score was assessed at screening, visit day 15 and visit day 45.
Mental and physical fatigue score
This score consists of 13 items of which 8 describe physical fatigue and 5 concern mental fatigue . The items were assessed in five categories with score values from 0 to 3: “better or much better than usual” (0), “same as usual” (1), “worse than usual” (2) and “much worse than usual” (3). This means, the higher the score, the worse the fatigue. The achievable score values were 0–15 for mental fatigue and 0–24 for physical fatigue. Items are on a relative scale to baseline, which has by definition a value of 1 (“same as usual”) for each item, resulting in a mental fatigue score of 5 and a physical fatigue score of 8. The score was assessed at day 2–13, day 15 and 45 of follow up.
For statistical analysis, a fixed model for repeated measures was fitted to the endpoints mental fatigue score and physical fatigue score. All results from day 2 to 15 were included to estimate the change from usual for each day assessed. The comparison of the NBMI treatment groups with the placebo group is based on the change from baseline on day 15.
A sensitivity analysis was performed for the results of the statistical examination based on non-parametric tests (Wilcoxon–Mann–Whitney test), because diagnostic tests and scrutiny of the model residuals hinted at a violation of the assumption of normality.
Neuromotoric test battery CATSYS
To record neurological symptoms caused by mercury intoxication objectively, a common computerized test system for measuring motoric skills was used (CATSYS). It assesses hand coordination, reaction time, postural tremor and postural stability (www.catsys.dk)  and was already successfully used to measure the effects of chronic mercury exposure [25, 28, 29]. The tremor intensity was statistically analyzed for differences between the treatment arms.
A set of performance indices available for the CATSYS was calculated. The resulting index values were referenced and compared to the performance of a large sound population given by the company . These indices evaluate the tests for tremor intensity and frequency, sway characteristics as much as the test for reaction time and the tests for hand coordination skills. The instructions for the CATSYS state, that an index value of 1 corresponds to the performance of a healthy population. Index results strongly deviating from 1 signify an abnormal performance.
Each participant received an individual instruction and demonstration of the tests by the investigator and was subsequently examined for about 15 min.
The finger-to-nose test detects tremor and can take on following values: Absent (0), slight (1) and moderate to severe (2) . The baseline was assessed at day 1 and compared to the results assessed at day 15 and 45 of follow up.
Only descriptive statistics have been used to present all efficacy variables. Continuous variables were summarized with sample size, median, minimum and maximum value. Categorical data are shown in frequency tables with number of subjects, frequency and percentage of occurrence. Individual data were presented in subject listings. When applicable, point estimates, together with their 95% confidence intervals have been presented.
The results of the two NBMI treatment groups were separately compared with the placebo group on day 15. To test for differences between the groups, the ANCOVA method, adjusting for baseline level, was used for parametric distributed characteristics, while the Wilcoxon–Mann–Whitney test was used for non-parametric results. Using the mixed model repeated measures method, a model was fitted to the daily measures of fatigue scores, mental and physical fatigue, including all results of days 2 to 15 to estimate the mean change from usual of fatigue at each day. The comparison between the groups treated with NBMI and the one treated with placebo is based on the change from usual on day 15. This change was individually calculated as the difference of baseline values and the values on day 15 or day 45 and the groups analyzed for distinctions in the change values. The initial model was fitted using an unstructured covariance between repeated measures. If convergence failed, the following covariance structures were tested in the given order until convergence was reached: Unstructured, Heterogeneous Toeplitz, Heterogeneous autoregressive, Toeplitz, Autoregressive.
All assessed data were included in the statistical summary and missing data were not imputed. All statistical analyses were performed using the version 9.4 of the SAS® software.
Adverse events were identified by spontaneous reports from the subjects, observations by the investigators or the medical personnel or through elicitation based on non-leading questions by the study personnel. All adverse events were recorded in the case report files along with a diagnosis, when available, or signs and symptoms, start and stop date and time, intensity (mild, moderate or severe), a causal relationship with NBMI (probable, possible or not related). If applicable, the action taken to handle symptoms and its outcome were recorded as well. Blood samples were collected for the examination of clinical chemistry and hematology and sent to a laboratory at screening and on days 1,6,15 and 45. On the same days, a full physical examination was done to control the vital signs and the general condition as well as the pulmonic, the cardiovascular, the abdominal and the neurological state of the participants.
The study protocol and study informed consent forms (ICF) were submitted to the Institutional Review Board (IRB) of the Universidad de San Francisco de Quito, Diego de Robles y Vía Interoceánica, Quito, Ecuador, for review. After requested revisions, the study protocol and ICF were approved in writing on 26 May 2015. The Regulatory Authorities required revisions and protocol (amendment 2) and ICF were therefore re-submitted to the IRB and approved on 27 Jul 2015.
The study protocol was amended once during the study (amendment 3) to protocol. The amendment, including ICF was submitted to the IRB and approved on 30 September 2015.
The study was performed in accordance with the ethical principles that have their origin in the Declaration of Helsinki and that are consistent with International Conference on Harmonisation/Good Clinical Practice and applicable regulatory requirements on Bioethics.