Vaginal microbiome transplantation in women with intractable bacterial vaginosis

We report the results of a first exploratory study testing the use of vaginal microbiome transplantation (VMT) from healthy donors as a therapeutic alternative for patients suffering from symptomatic, intractable and recurrent bacterial vaginosis (ClinicalTrials.gov NCT02236429). In our case series, five patients were treated, and in four of them VMT was associated with full long-term remission until the end of follow-up at 5–21 months after VMT, defined as marked improvement of symptoms, Amsel criteria, microscopic vaginal fluid appearance and reconstitution of a Lactobacillus-dominated vaginal microbiome. One patient presented with incomplete remission in clinical and laboratory features. No adverse effects were observed in any of the five women. Notably, remission in three patients necessitated repeated VMT, including a donor change in one patient, to elicit a long-standing clinical response. The therapeutic efficacy of VMT in women with intractable and recurrent bacterial vaginosis should be further determined in randomized, placebo-controlled clinical trials. This first-in-human study assessed the feasibility of vaginal microbiome transplantation from healthy donors as a therapeutic alternative for patients suffering from symptomatic, intractable and recurrent bacterial vaginosis.

severe end of the BV spectrum. Fecal microbiome transplantation (FMT), in which feces from healthy donors are introduced into recipients' intestines to replace their disease-associated microbiome, has recently been successfully used in treating severe and recurrent Clostridium difficile infection 23 . Although gastrointestinal microbiome interventions may offer a different ecological scenario than those related to a dysbiotic vaginal microbiome, we hypothesized that a similar use of VMT might be beneficial in treating the most severe cases of recurrent and antibiotics-nonresponsive BV.
Five patients were recruited (aged 27-47 years and referred to as patients A-E; Extended Data Fig. 1). All suffered from intractable BV, defined as four or more symptomatic episodes of BV during the previous year 19 , relapsing after repeated, prolonged and diverse antibiotic attempts requiring continuous maintenance antibiotic treatment to remain symptom free. All patients reported a substantial negative impact of BV symptoms on their quality of life, including devastating consequences to their relationships, sexual intimacy and self-esteem. All five patients were otherwise healthy. Patient screening, exclusion criteria and the consent process are described in the Methods. The three donors were premenopausal, healthy volunteers, aged 35-48 years (donors 1-3; Extended Data Fig. 1), who did not report having BV in the last 5 years or any history of recurrent BV. Donor selection and screening are detailed in the Methods. Repeated communication between the lead physician and the donors ensured that the behavioral requirements (e.g., abstinence from sexual activity for 1 week before donation in the sexually active donor) were strictly followed.
Before transplantation, all patients were treated with an intravaginal antibiotic regimen 16 that previously resulted in a longer symptom-free period, which consisted of 5 g clindamycin cream (2%) for 7 d (recipients B, C and E) or 5 g metronidazole gel (0.75%) for 5 d (recipients A and D). VMT was performed 1 week after completion of antibiotic treatment 24 . During the procedure, vaginal fluid for transplantation was collected from the donors starting from the seventh day of the menstrual cycle (Methods) and taken from the upper half of the vagina and cervical fornices, while avoiding the cervix. The collected discharge was evaluated by pH and microscopy, diluted with 1 ml of sterile saline and transferred to the recipient's posterior fornix, without the use of a speculum (Fig. 1a). VMT was completed within no more than 60 min of sample collection and was performed at any stage during the recipient's menstrual cycle, except during menstruation. After the first VMT, repeat VMTs were performed in cases of symptom recurrence or with reappearance of one or more positive Amsel criteria during
Four patients (patients A-D) had long-lasting improvements in their Amsel scores (Fig. 1b), microscopic vaginal fluid appearance (Fig. 1c,d), whiff test, discharge and vaginal fluid pH (Fig. 1d   and 2, respectively. Both reported immediate clinical improvement, with the disappearance of odor within 1 week of transplantation, and a gradual decrease in discharge, resulting in no symptoms 1 month after VMT. In both patients, normalization of all Amsel criteria as well as normal Lactobacillus-dominated microscopic appearance was documented 1 week after transplantation and persisted on follow-up examinations (11.5 months in patient A and 5.5 months in patient B). Patient C received the microbiome of donor 1. She reported an improvement of symptoms after VMT and became BV negative according to Amsel criteria, but her microscopic findings were consistent with persistence of BV. She therefore underwent a repeat VMT from the same donor (donor 1) without preceding antibiotic treatment. For 4 months, the patient reported an improvement of symptoms and BV status was negative according to Amsel criteria. However, 4.5 months after the first VMT, she experienced a recurrence of odor, positive Amsel criteria and a BV microbiome appearance on microscopy, all consistent with recurring BV. She, therefore, underwent a third VMT, this time using a sample from a different donor (donor 3) after vaginal antibiotic treatment. After this VMT, she reported complete resolution of symptoms; Amsel criteria were normalized; and microscopy showed a normal Lactobacillus-dominated appearance for 11 months of follow-up. Patients D and E likewise had a fluctuant course. After a first VMT from donor 1, patient D experienced a recurrence of symptoms, positive Amsel criteria and microscopic findings consistent with BV. She underwent a second VMT from the same donor (donor 1), after which she reported clinical improvement of symptoms and was BV negative according to Amsel criteria. However, she exhibited an intermediate vaginal microbial appearance on microscopy and therefore underwent a third VMT from the same donor (donor 1), after which she reported clinical improvement with the disappearance of odor and improvement of discharge, associated with negative Amsel criteria and a normal Lactobacillus-dominated appearance on microscopy. On evaluation 21 months after the third transplant, the patient reported no recurrences, had negative Amsel criteria and exhibited normal microscopy. After VMT from donor 2, patient E reported a partial symptomatic improvement, associated with negative Amsel criteria and a normal Lactobacillus-dominated appearance on microscopy, for 4 weeks of follow-up. She then took systemic antibiotics for pharyngitis and soon after reported a recurrence of odor, accompanied by positive Amsel criteria and BV-characteristic microscopic appearance. She underwent a repeat VMT from the same donor (donor 2), resulting in the normalization of all Amsel criteria and improvement of her microscopic vaginal appearance to an intermediate microbiome configuration, coupled with partial symptomatic improvement, for 6.5 months of follow-up. To characterize the genus-level changes associated with VMT, we sequenced all donors' and recipients' vaginal microbiome samples using 16S ribosomal DNA (rDNA) sequencing (Methods). Interestingly, healthy microbiomes clustered differently from the microbiomes of BV-diagnosed patients (Extended Data Fig. 4a) after applying principal-coordinates analysis (PCoA) with UniFrac distances 25 . Using Bray-Curtis (BC) dissimilarity, we followed recipients' microbiomes before and after VMT and observed a rapid change in the composition of all microbiomes after VMT, correlated with recovery in all of the Amsel criteria (Extended Data Fig. 4b).
To study the effect of VMT on vaginal microbiome species-level composition and metagenomic function, all donors' and recipients' samples underwent shotgun metagenomic sequencing. As expected, the microbiomes of donors and recipients were found in two distinct clusters using PCA (Fig. 2a). The effect of VMT on global microbiome composition over the follow-up period was assessed by BC dissimilarity on the species level, as compared to patients' baseline BV configuration. Four of five VMT recipients exhibited a drastically changed microbiome composition already at the first month after VMT, which correlated with a notable recovery of their Amsel criteria (Fig. 2b) as well as with every discrete clinical criterion (Extended Data Fig. 5a). Patient C experienced the same trend, only after the third and successful VMT (Fig. 2b). The post-VMT microbiome of one patient (patient E) relapsed to her baseline microbiome BV composition after failure of the first VMT (Fig. 2b). However, the repeat successful VMTs in this patient induced a distinctively different configuration, mirrored by a marked specieslevel BC distance from baseline, similarly to the post-VMT trend observed with the other four patients after a successful VMT (Fig. 2b). In four of the five VMT recipients, the vaginal microbiome configuration remained distinct from the baseline BV configuration over a period of 5-21 months after a successful VMT (Fig. 2c). Notably, the post-VMT vaginal microbiome composition became significantly more similar to that of the collective donor vaginal microbiome configuration, as compared to the corresponding similarity between the pre-VMT and donor configurations (Euclidean distances, P = 0.0012; Fig. 2c and Extended Data Fig. 5b). This similarity was present after a successful VMT and through the followup period (Fig. 2d). Notably, the current preliminary case series is underpowered to statistically test a person-specific donor contribution to a recipient's specific clinical features or microbiome configuration after VMT. Larger future cohorts may enable better resolution or, alternatively, demonstrate that distinctions can be made between only a 'healthy' versus a BV microbiome configuration.
This post-VMT compositional change was mostly dominated by an expansion in members of the Lactobacillus genus, combined with a decrease in members of the Bifidobacterium genus, closely related to the Gardnerella genus. (The reference that was used for taxonomic annotations classifies Gardnerella genus and Gardnerella vaginalis specie as Bifidobacterium and Bifidobacterium vaginale, accordingly; Fig. 2e and Extended Data Fig. 5c.) Other genera, including Fannyhessea and Prevotella, were reduced upon successful VMTinduced remission of BV (Extended Data Fig. 5d). We further used species-level PCA in reducing the complex microbiome dimensionality (Extended Data Fig. 5e). Indeed, the PCA clustered the samples into a BV cluster, containing mostly samples with one or more Amsel-diagnosed BV features, and a healthy cluster, with mostly no diagnosed clinical features (Extended Data Fig. 5e). We applied a k-means algorithm (k = 2), using the coordinates of the first and second PCs, to define the two clusters that were visually identified (Extended Data Fig. 5e). We then calculated the purity score for each Amsel criteria score division. Considering the purity scores, we classified our samples into two groups according to their Amsel score (i.e., the first group comprised all samples having Amsel criteria = 0 and the second group comprised all samples with Amsel criteria > 0). To see whether the groups were indeed different, we conducted a permutational analysis of variance test using the BC dissimilarity matrix (P < 0.05). The difference between the two clusters could be explained by the relative levels of Bifidobacterium and Lactobacillus genera in each sample (Extended Data Fig. 5f). The most dominant features that contributed to the change in the first PC, which differentiated between the clusters, consisted mostly of Lactobacillus crispatus specie in the healthy cluster and Bifidobacterium vaginale in the BV cluster (Fig. 2f), demonstrating the 15 overall most PC1influential taxa, as represented in each vaginal microbiome configuration). Interestingly, recipient E, who was the only partial clinical responder, featured a different dominant post-VMT lactobacillus strain (Lactobacillus gasseri), that was not one of the top 15 PC1influential strains in the other four VMT recipients).
Functional microbiome changes after VMT, as assessed using the Kyoto Encyclopedia of Genes and Genomes (KEGG), revealed two distinct functional clusters that separated the BV microbiome from the healthy one (Fig. 2g), corresponding to the taxonomic differences noted between these conditions (Fig. 2h). Upon recipient follow-up, functional BC distances from the baseline microbiome correlated with a decrease in Amsel criteria (Extended Data Fig. 6a).
Changes in the functional potential of the microbiome shifted at the time of VMT (Extended Data Fig. 6b) and remained unaltered over the follow-up period (Extended Data Fig. 6c). Similarly to the taxonomic analysis, the current preliminary case series was underpowered to statistically test person-specific donor and recipient similarities in functional microbiome characteristics, and these could only cluster as BV, healthy donor and post-VMT groups. An alternative functional analysis using Gene Ontology (GO) terms demonstrated that the post-VMT BC distance from the baseline likewise correlated with the decrease in Amsel criteria in three patients (patients A-C; Extended Data Fig. 7a), whereas in two patients (patients D and E) the distances remained unchanged. Nonetheless, a clear PCA cluster could be observed (Extended Data Fig. 7b) between the BV and healthy microbiome configurations, and these could be clearly linked to the different taxonomic composition of the BV and healthy clinical states (Extended Data Fig. 7c). The most dominant GO terms remained stable throughout VMT and the follow-up period, potentially explaining the low BC changes we observed using this analytical method (Extended Data Fig. 7d), whereas the second PC exhibited a substantial change in GO term signatures over the course of the follow-up period (Extended Data Fig. 7e).
Collectively, we report the feasibility of using VMT as a long-term treatment for recurrent, antibiotics-nonresponsive and intractable BV. Although we did not observe adverse effects associated with VMT in this study, we cannot completely exclude potential risks associated with any microbiome transfer procedure. The transplant of antimicrobial-resistant microbes has been reported in immunocompromised patients undergoing FMT 26 , and the long-term consequences of VMT remain unknown. Gynecologic and obstetric complications, however unlikely, are also possible. Additionally, the risks of unintended pregnancy due to the transfer of sperm or the transfer of undetected pathogens with the vaginal fluid are not negligible. Ιn our study, donor selection followed stringent criteria to minimize the risks, yet these criteria may not be applicable in other settings. The use of contraception by recipients as a mandatory criterion in future studies, associated with the development of a 'vaginal fluid bank' in which samples from suitable donors will be kept for a period allowing for repeated screening, and verification of the absence of HIV sero-conversion before VMT, may be recommended. Finally, BV may be asymptomatic or readily treatable with antibiotics in most women, and therefore VMT should be considered only in cases of multiple treatment failures and substantial disruption of the patient's quality of life due to chronic and intractable symptoms. Although all patients enrolled in this exploratory study benefited from the procedure, the efficacy of VMT as a treatment in intractable BV needs to be determined in randomized, placebo-controlled trials.

Online content
Any methods, additional references, Nature Research reporting summaries, source data, statements of code and data availability and associated accession codes are available at https://doi.org/10.1038/ s41591-019-0600-6.

Methods
Human study cohort. This study was conducted at the Hadassah Medical Center in Jerusalem, Israel. All participants provided written informed consent. The research protocol was approved by the ethics committee at the Hadassah Medical Center (HMO-0667-13) and the Weizmann Institute of Science (603-1, 680-1) with ClinicalTrials.gov ID NCT02236429.
Bacterial vaginosis. BV was diagnosed by the Amsel criteria, requiring three of the four following symptoms or signs: homogeneous, thin, white discharge; pH >4.5; a fishy odor of vaginal discharge before or after addition of 10% potassium hydroxide (i.e., the whiff test); and >20% vaginal epithelial cells studded with adherent coccobacilli (clue cells) on microscopic examination 16,27 . On microscopy, bacterial microbiome appearance was defined as normal (lactobacilli dominated), BV (coccid-bacillary dominated) or intermediate, as quantified by using the Hay-Ison criteria 28 .
Recipients. Inclusion criteria. Recipients were included if they were aged 18-50 years with recurrent BV, defined as ≥4 symptomatic episodes of BV during the previous year, and required maintenance antibiotic treatment (twice weekly) to remain symptom free or if they experienced recurrence of BV in ≤2 months after treatment, with a documented history of recurrent BV. Exclusion criteria included pregnancy or planned pregnancy in the upcoming year or infection with hepatitis B, hepatitis C, HIV or syphilis. Study candidates underwent screening for cervicovaginal infection with Chlamydia trachomatis, Neisseria gonorrhea, Mycoplasma genitalium or Trichomonas vaginalis, using a PCR assay. Any patient who presented a positive result for any of these infections received the standard recommended treatment, with a documented negative assay result deemed mandatory for inclusion in the study. All patients underwent a cervical cytology screening test (Pap test) and PCR-based screening for human papilloma virus (HPV). In the case of an abnormal cytology test or a positive HPV test, patients were referred for colposcopy. In addition, vaginal cultures for yeast and bacteria (streptococci groups A, B, C and G), urine cultures, urinalysis, and serology analysis for HIV, hepatitis A, B and C, Treponema pallidum, herpes viruses and cytomegalovirus (CMV) were performed in all cases. Donors' long-term medical and sexual history was familiar to the lead clinician. Two were non-sexually active for 8 years or more; one was engaged in a 25-year monogamous relationship. Donors had a negative history of vaginal symptoms and underwent an examination to verify the absence of BV and other vaginitis, using a thorough history, gynecologic exam, vaginal fluid microscopic assessment, cultures and PCR assays. All donors answered a questionnaire that is used at our blood bank (see below) to screen for possible risk factors for acquiring an infection that we potentially missed using PCR assays, cultures and serology. Donors were screened for potentially important infections, including group B Streptococcus (GBS) and CMV, so as not to expose women who were GBS or CMV negative to potential future complications of pregnancy and delivery caused by GBS or CMV. Before vaginal fluid collection, it was explicitly verified with the donors that they did not have sexual intercourse in the week preceding the intervention.

Donor history questionnaire.
1. Are you generally healthy? 2. Do you suffer from any health problem? If you do, please specify. 3. Have you taken any medication during the past month? If so, please indicate each one. 4. Have you traveled abroad during the past 12 months? If so, please specify where and when. 5. Did you get a tattoo/ear or body piercing/accidental needlestick during the last 6 months? 6. Were you bitten by an animal in the last few months? 7. Did you suffer from hepatitis? 8. Did you live with a person who has hepatitis in the past 6 months? 9. Did you suffer from tuberculosis during the past two years? 10. Did you receive antibiotic treatment for a sexually transmitted infection (e.g., Chlamydia, gonorrhea, trichomonas) in the past year? 11. Did you live in a malaria-infected area or suffer from malaria? 12. Did you undergo any surgery? If yes, please specify which? When? 13. Did you receive a blood transfusion? If so, when? 14. Were you pregnant or delivered during the past 6 months?
If an answer to any of the following questions is positive, we ask that you do not participate as a donor: Consent process. All patients underwent a thorough and detailed consent process, including being informed of the hazards potentially mitigated by this experimental approach. They received a highly detailed explanation of the study procedures. They were also informed about the limited ability to completely predict future gynecologic and obstetric complications; the possible transfer of infectious agents, including those that cannot be screened for; as well as the small but non-negligible risk of inadvertent sperm transfer. For patients who did not use an intrauterine device, the use of hormonal contraception was recommended. A time interval of 7-14 d was set after the detailed explanation was given and before deciding on participation to let the patient consider all of her options.
VMT sample collection and transfer. The flat, broad part of Ayre's spatula was used for sample collection. This device, originally designed for vaginal sample collection for the Pap test, does not absorb the fluid, is shaped for vaginal use and does not wound the vagina. In parallel to VMT sample collection, samples were taken, using the same technique, for molecular analysis, using the ESwab Multiple Specimen Collection and Transport System (COPAN) and stored at −80 °C. After VMT, recipients were instructed to avoid sexual intercourse for 1 month; avoid bathing (in a bath, hot tub, swimming pool, etc.) for 1 week; avoid douching, intravaginal medications and systemic antibiotics for 1 month; and avoid probiotics for the entire follow-up period of 1 year.
Follow-up after VMT. At each appointment, patients underwent a gynecologic examination that consisted of an evaluation by the Amsel criteria including microscopy of vaginal discharge. In the case of an abnormal cytology test or a positive HPV test at recruitment (four patients), patients underwent colposcopy before VMT and were followed as recommended by American Society for Colposcopy and Cervical Pathology (ASCCP) guidelines, including colposcopy, HPV test and cytology screening. One patient with a normal cytology test and a negative HPV test had repeated cytology screening at the end of 1 year of follow-up and was instructed to continue follow-up as recommended by ASCCP guidelines. As recipients continued to have sexual intercourse after VMT, an infection possibly detected in the time after VMT could not necessarily be attributed to the VMT and was therefore not routinely assessed.
16S sequencing and analysis. For 16S amplicon sequencing, PCR amplification was performed of the 16S rDNA gene and subsequently sequenced using 500-bp paired-end sequencing (Illumina MiSeq). Amplicons spanning variable region 4 (V4) of the 16S rDNA gene were generated by using the following barcoded primers: Fwd 5 15 F, A AT GA TA CG GC GA CC AC CG AG AT CT AC AC TA TG GT AA-TT GT GTGCCAGCMGCCGCGGTAA; Rev 806R, C AA GCAGAAGACGGCATA CGAGATXXXXXXXXXXXXAGTCAGTCAGCCGGACTACHVGGGTWTCTA AT; where X represents a barcode base. W = A or T; M = A or C; H = A or C or T. The reads were then processed using the QIIME 2.2019-1 pipeline 29 . In brief, fastq quality files and a mapping file indicating the barcode sequence corresponding to each sample were used as inputs; reads were split by sample according to the barcode, then they were denoised by the dada2 plugin and taxonomic classification was performed using the Greengenes database and the naive Bayes pretrained QIIME2 classifier. Rarefaction was used to exclude samples with insufficient count of reads per sample. For beta diversity, UniFrac measurements were plotted on the basis of 70,000 reads per sample.
Shotgun metagenomics sequencing and processing. Genomic DNA was purified using a Purelink Microbiome DNA purification kit (Invitrogen) optimized for a Tecan automated platform. For shotgun sequencing, Illumina libraries were prepared using a Nextera DNA Sample Prep kit (Illumina, FC-121-1031), according to the manufacturer's protocol, and sequenced on the Illumina NextSeq platform with a read length of 80 bp. We then used Illumina's bcl2fastq script to make the fastq files. Host reads were removed using KneadData with default parameters and the hg19 reference. Taxonomic assignment was performed with Kraken2 30 , using this recommended prebuilt index database 31 . From the resulting table, we extracted the counts for both genus and species levels separately. We then subsampled the count tables so each sample had 100,000 reads in total. Three samples that did not reach 100,000 reads were excluded from the taxonomic analysis. On top, we filtered out all bacteria whose total abundance after rarefaction was 10 -4 . In all clinical-microbiome distance correlations, only time points that included both clinical and microbiome measurements were used. Functional annotation was performed using the Humann2 32 pipeline, with the same input reads we used for the taxonomic analysis. The output was normalized to counts per million, combined and annotated to KEGG and GO terms using Humann's built-in scripts (humann2_regroup_table with UniRef90ko and UniRef90go, respectively, together with humann2_rename_table). These datasets were later filtered to remove all genes and terms for which the abundance was <10 -3 .
Statistical analysis. Paired t-tests were used to compare baseline and post-VMT distances to the centroid of the three donors. The distances were computed on all PCs. PCA was performed using the scikit-learn package in Python, as were all the analyses, after performing a log transformation in all cases. Figure 2f shows the 15 leading loadings (in absolute value) for the first and second PCs. k-means clustering was also performed using scikit-learn, for 100 iterations with random_ state = iteration's index.
Permutation tests were performed in the following manner. Under the null hypothesis that the microbiome profiles of samples originating from both groups have the same distribution, we let μ 0 denote the mean of pairwise dissimilarity between the original groups. For i = 1 to 10 5 , we shuffled the labels of the groups stratifying the permutations so that labels were switched only within the same subject's samples. We then denoted the mean of pairwise dissimilarity of the relabeled groups by μ 1 . The probability of the null hypothesis is p ¼ 10 5  I For example, we counted each iteration that had a higher mean of pairwise dissimilarity, and the P value is the sum of these divided by the number of iterations.
Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The sequencing data has been deposited at the European Nucleotide Archive with accession number PRJEB34085. All other requests for raw and analyzed data, generated as part of a clinical trial, will be promptly reviewed by the Hadassah-Hebrew University Medical Center and the Weizmann Institute of Science to verify whether the request is subject to intellectual property confidential obligations or affects patient confidentiality.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.

Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A list of figures that have associated raw data -A description of any restrictions on data availability ENA : PRJEB34085 Field-specific reporting Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.

Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
October 2018 the donors that they did not have intercourse in the week preceding the intervention.

Ethics oversight
Hadassah-Hebrew University Medical Center; Weizmann Institute of Science Note that full information on the approval of the study protocol must also be provided in the manuscript.