Serena Porcari
2026-01-28 00:00:00
Study design and approvals
The TACITO trial was an investigator-initiated, randomized, double-blind, placebo-controlled phase 2a clinical trial that aimed to evaluate whether donor FMT from complete ICI responders is effective in improving response to combined first-line therapy with pembrolizumab and axitinib in patients with mRCC.
The study was conducted in accordance with the Declaration of Helsinki and International Conference on the Harmonization of Good Clinical Practice guidelines as well as in compliance with local and institutional regulations. The study was approved by the institutional review board (IRB)/local ethics committee (ID: 2664) and was prospectively registered at ClinicalTrials.gov (registration identifier: NCT04758507, registration date: 11 February 2021). All enrolled patients gave their written informed consent to participate in the study. This study was conducted by following CONSORT guidelines37, and a CONSORT checklist is provided in Supplementary Table 1.
The full study protocol is available in the Supplementary Information file.
Study population
Patients were enrolled in the Fondazione Policlinico Gemelli IRCCS, a tertiary care center and academic hospital in Rome, Italy, and were either screened from our oncology outpatient clinic or referred to our center from seven other Italian hospitals, all referral centers for RCC. The inclusion criteria were as follows: histologically confirmed RCC; metastatic disease; radiological assessment within 8 weeks before enrollment; patient eligible for therapy with ICIs for mRCC or started within 8 weeks; ability to provide written informed consent; and ability to be compliant with the scheduled procedures.
We excluded patients for the following reasons: major comorbidities; concomitant gastrointestinal or autoimmune disorders or HIV, HBV or HCV infection; continuative corticosteroid therapy; previous treatment with systemic immunosuppressants or immunomodulatory drugs; and antibiotic therapy within 4 weeks prior to enrollment.
Predefined withdrawal criteria before the start of treatments (checked at the pretreatment evaluation visit, described below) were as follows: voluntary withdrawal of the patient; absence of the planned eligibility criteria at the pretreatment evaluation visit; and actual provision of the first treatment (FMT or placebo) later than 8 weeks from the start of ICI therapy. We predefined this criterion as a late provision of the first treatment that could affect the efficacy of FMT. The delay of further treatments was not considered a criterion to withdraw patients from the study as the donor microbiome can engraft the recipient gut up to 6 months, based on available evidence38. Patients who experienced a disease progression during the study procedures were discontinued from further treatments. Moreover, patients who withdrew their consent to participate in the study at any time were discontinued from the trial.
The last enrolled patient received his first treatment on 16 December 2023 and had his last follow-up visit on 5 December 2024.
Two of the 25 patients randomized in the d-FMT arm did not receive the first treatment as the timespan between the scheduled procedure and the start of their cancer therapy was longer than 8 weeks due to unavailability of donor stools. Three of the 25 patients randomized in the p-FMT arm did not receive the first treatment for the following reasons: one patient withdrew consent before the scheduled procedure; one patient had no metastatic disease at internal reevaluation of the baseline computed tomography scan; and one patient was not treated within 8 weeks from the start of cancer therapy, because of his logistical unavailability to reach our center.
Analysis sets
ITT set. This set was used for sensitivity analyses and included all randomized patients, regardless of study treatment.
FAS. This set was used for efficacy analyses (primary and secondary efficacy outcomes) and included all randomized patients who received at least the first treatment (regardless of completion status).
Per-protocol set. This set was used for sensitivity analyses and included all randomized patients who received all assigned treatments for which they were eligible (so, unless disease progression, consent withrawal or other conditions for study discontinuation occurred) within the planned schedule (regardless of their number), without delays.
Safety set. This set includes all randomized patients, regardless of study treatment. Only the TRAEs and the adverse events of special interest, which included gastrontestinal toxicities, have been collected prospectively. Incidence of treatment interruption or discontinuation related to the axitinib and/or pembrolizumab as well as the incidence of grade 3–5 adverse events in each treatment arm have been collected retrospectively.
Moreover, we conducted a post hoc analysis by excluding patients with favorable-prognosis disease based on the IMDC score.
Study endpoints
The primary efficacy endpoint was 12-month PFS, defined as the number of participants free from tumor progression, as assessed by Response Evaluation Criteria in Solids Tumors (RECIST) version 1.1, 12 months after randomization39, in the FAS population.
Secondary efficacy endpoints were median PFS, median overall survival and ORR in the FAS population, safety and microbiome changes after treatments.
Safety was evaluated in the per-protocol and FAS populations.
PFS and ORR were determined using RECIST version 1.1, as assessed by local investigators. PFS was defined as the time from randomization until disease progression or death, whichever occurred first. Overall survival was evaluated from the date of randomization to death from any cause or the last contact.
ORR was defined as the proportion of patients with a partial or complete response.
Safety was assessed as the incidence, nature and severity of TRAEs, recorded and classified according to Common Terminology Criteria for Adverse Events (CTCAE) version 4.0.
Microbiome changes after treatments were defined as the number of participants with significant increase in α-diversity (assessed by Shannon index) and β-diversity (assessed by Bray–Curtis dissimilarity) of their gut microbiota after treatments, compared with baseline.
Study procedures
Screening, baseline and pretreatment assessments
Screening, baseline and pretreatment assessments were all conducted in combination by physicians from the Gastroenterology Unit and from the Oncology Unit.
During the screening visit, after a thorough discussion of the study details with the investigators, potentially eligible patients who were willing to participate in the study signed the informed consent. Then, we evaluated the patient and reviewed inclusion and exclusion criteria to ensure the eligibility of the patient for the trial.
The baseline assessment included the collection of patient medical history, including the following data: age, gender, date of cancer diagnosis, nephrectomy status and date, histology, Fuhrman grade, diagnosis of metastatic disease and site of metastases, Karnofsky scale, IMDC score, complete blood count and starting date of cancer therapy. Moreover, all patients underwent a full clinical examination. At the end of the baseline visit, enrolled patients were randomized to study treatments.
The pretreatment assessment occurred within 2–5 days before the first scheduled experimental procedure (FMT or placebo). Here, investigators reviewed the eligibility criteria to check their actual validity as well as the time distance between the start of ICI therapy and the first treatment (to be no longer than 8 weeks).
Selection of stool donors
Stool donors were selected within the historical cohort of our Oncology Unit among patients with advanced RCC who had experienced a complete response to PD-1 inhibitors (pembrolizumab or nivolumab).
Once identified, available patients from this cohort were screened according to protocols recommended by international guidelines, based on a multilevel framework32,40,41. First, donor candidates underwent a specific questionnaire aimed at addressing the following: known history or lifestyle-related risk factors for potentially communicable diseases (for example, drug addiction or promiscuous sexual behavior); recent (32,40,41. At the time of each donation, the chosen patients underwent a further questionnaire to screen for any recent acute digestive disease, newly contracted infectious diseases or other potentially harmful situations (for example, risky sexual contacts) and a nasopharyngeal swab for SARS-CoV-2 (after March 2020)32,40,41. Moreover, each stool donation provided by qualified donors underwent direct stool testing, which included a culture assay for detection of multidrug-resistant bacteria, a rapid molecular assay for common intestinal pathogens (Seegene, RT–PCR Allplex Gastrointestinal Panel Assay) and a stool molecular assay for SARS-CoV2 (since March 2020)32,40,41. An extended description of the donor screening is available in the study protocol. Selected donors were also asked to provide stool samples for microbiome analysis every 6 months.
Five patients with long-term complete response to ICI were identified as candidate donors.
Three donors did not pass the screening process (two for chronic use of proton pump inhibitors and another one for history of previous colorectal surgery for diverticulitis), whereas two of them were eligible to donate.
Donor 1 was a 57-year-old man who underwent radical nephrectomy and retrocaval lymph node resection for a clear cell RCC at original pT3a pN1 M0 disease stage. After 4 months, for disease recurrence with more than 60 bilateral pulmonary metastases (the largest of 14 mm in size), he started first-line systemic therapy with nivolumab + ipilimumab, achieving a complete response that is still lasting after 6 years of therapy (that is, May 2025). He provided the vast majority (n = 56/57, 98%) of stool aliquots. The other donor who passed the screening process (donor 2) was a 52-year-old woman who developed lung and pancreatic metastases 2 years after nephrectomy for a clear cell RCC (pT2, grade 2) and received third-line therapy with nivolumab after disease progression to sunitinib for 3 years and everolimus for subsequent 3 years. Nivolumab led to a complete disease response, and, as of today (that is, May 2025, 9 years after), she is still under therapy with no evidence of target lesions.
Manufacturing of fecal aliquots and FMT capsules
For each donation, feces was collected by the donor in the morning and rapidly transported to our hospital in a refrigerated bag (within 4 hours from defecation). After the completion of the questionnaire already described, stool donations were brought to the FMT laboratory of our center (Microbiology Unit) for direct stool testing. Immediately after this step, they were manufactured as fecal aliquots for colonoscopic FMT and/or as FMT capsules, based on specific needs related to planned treatments, following recommendations by international guidelines32,40,41. All manufactured aliquots or capsules were quarantined for 24–48 hours while waiting for results of the direct stool testing, and positive donations were then discarded. Also, stool donations less than 60 g were considered insufficient to prepare a stool aliquot or a capsule set and were discarded42. All procedures were performed within a biological safety cabinet (Biosafety Level 2). Using specific strainer bags (Seward), the fecal material was first diluted with nearly 200 ml of sterile saline (0.9%) for fecal aliquots and nearly 100 ml of sterile saline for FMT capsules, and glycerol (Monico S.p.A.) was added up to a final concentration of 10% to cryopreserve bacteria at −80 °C. Then, this suspension was homogenized automatically through a Stomacher 400 homogenizer (Seward), which allowed a simultaneous filtration of the solution. The homogenization program chosen was 260 rev min−1 for 1 minute43. Further filtration and purification of the fecal suspension were performed using sterile gauzes and a funnel. The deriving solution was blended and, after the supernatant was strained, transferred into a sterile flask.
For the manufacturing of colonoscopic FMT aliquots, the fecal suspension was divided, in ready-to-use aliquots of at least 60 g and nearly 200 ml, as previously described43, and aliquots were frozen at −80 °C up to 6 months. Eligible aliquots were thawed in a warm (37 °C) water bath on the day of fecal infusion.
For the manufacturing of FMT capsules, the fecal suspension was poured into a sterile container ready for capsule production. Specifically, size 1 capsules were filled with 0.420 ml of fecal suspension and, then, encapsulated with size 00 capsules (Farmalabor), with nearly 150 capsules for each stool donation44,45. Capsules were quickly stored at −80 °C, in plastic boxes, for up to 6 months.
Each suspension bottle or capsule box was labeled with a unique barcode of the corresponding donor, and with the date of collection/manipulation, to assure its complete traceability, as recommended by international guidelines40,41. The personnel involved in the manufacturing of FMT were not involved in any part of the clinical trial and had no contact with patients.
Colonoscopic FMT
All colonoscopy procedures were performed in the morning, under sedation. Before colonoscopy, patients in both groups underwent bowel cleansing with 4 l of Macrogol (SELG ESSE) the day before the procedure. All procedures were performed by two expert endoscopists (G.I. and G.C.) using pediatric colonoscopes and carbon dioxide insufflation. Both fecal aliquots and placebo aliquots were delivered through the operative channel of the scope after reaching the cecum or the more proximal point of the large bowel, using 50-ml syringes filled with the infusate during colonoscopy. After the procedures, patients were monitored in the recovery room of our endoscopy center for nearly 3 hours. Patients were requested to fast for at least 6 hours after the procedure and were allowed to have a light meal thereafter.
Capsulized FMT
Patients were scheduled to receive capsulized FMT, respectively, 12 weeks (±2) and 24 weeks (±2) after the first treatment. Delay in receiving the scheduled treatments longer than 2 weeks was considered a protocol deviation. Patients received a styrofoam box, filled with dry ice and 10 capsule containers. Each container included 15 capsules. Patients were instructed to put capsule containers at −20 °C (the home freezer) and to take five capsules three times a day (at least 1 hour after a meal) for 10 days, for a total of 150 capsules. Originally, we planned to give 120 capsules for each treatment cycle, but we slightly increased the number of capsules as FMT capsules needed to be filled with a slightly lower quantity of fecal material to avoid them becoming damp and then potentially identifiable by the participants. Patients were asked to report the capsule assumption and to bring back capsule containers after the end of therapy, to monitor the assumption of capsules.
Follow-up assessments
Follow-up visits were performed in combination by physicians from the Gastroenterology Unit and from the Oncology Unit. Follow-up visits were scheduled at week 1, week 4, week 12, week 24 and week 52 after the first treatment, respectively. At each follow-up visit, the investigators evaluated the efficacy and safety of treatments, as described in the ‘Study endpoints’ subsection. Dates and status of disease progression, as well as date and type of best response, were recorded. Unscheduled follow-up visits were offered if requested by the patients.
Stool collection, DNA extraction and shotgun metagenomic sequencing
Samples were collected during the pretreatment visits and at the planned follow-up visits by using a stool collector with a DNA/RNA Shield buffer (Zymo Research), brought directly by patients to our center in a refrigerated box within 6 hours from collection and then stored at −80 °C for up to 36 months before being shipped in dry ice to the next-generation sequencing facility of the Department of Cellular, Computational and Integrative Biology of the University of Trento (Trento, Italy) for DNA extraction and sequencing. DNA extraction consisted of sample homogenization followed by DNA isolation with the PowerSoil Pro DNA Isolation Kit (Qiagen). Metagenomic sequencing libraries were prepared with the Nextera DNA Library Preparation Kit (Illumina) and sequenced on the Illumina NovaSeq 6000 platform with a target depth of 7.5 Gbp for patient samples (n = 247) and 15 Gbp for donor samples (n = 9).
Metagenome preprocessing and microbiome profiling
Metagenomes underwent quality control with our previously validated preprocessing pipeline (available at https://github.com/SegataLab/preprocessing). This consisted of removal of low-quality reads (Q Q > 2 ambiguous bases or 7.5 Gbp (approximately 50 million reads). These were then used as input in MetaPhlAn 4 (version 4.1)11,12 using the marker genes database vJun23_202307 to generate taxonomic profiles at species-level genome bin (SGB) resolution for each sample.
Assessment of donor strain engraftment
DoSER was defined as the number of strains (that is, specific genomic variants within species that can be profiled at the single-nucleotide level from metagenomes) of the donor that are present in each recipient’s sample after the FMT. In the case of d-FMT samples of individuals who had experienced donations from multiple aliquots, the pool of strains present in any of the received aliquots is considered in this estimation. For example, a DoSER of 50% in a sample means that half of the strains that are detectable in the administered donor’s fecal samples were also observed in that specific post-FMT recipient sample. As the fraction of a patient’s microbiome strain pool is also present in a different patient’s microbiome, the DoSER inference can be applied to any pair of samples, including samples from unrelated individuals not linked by FMT. The DoSER on unrelated individuals (for example, the baseline of the d-FMT group) is used here to assess the baseline strain sharing in a target population (as some strains might be clonally spreading in a population, and there can be direct or indirect social interactions within populations leading to strain transmission46); it can be applied also in the case of p-FMT patients to verify the absence of a donor’s effects in this arm. To detect engraftment of donor strains, we rely on our strain-sharing inference method previously reported31 plus recent improvements to the pipeline46,47. In brief, this consists of building strain-level phylogenies for all SGBs detected in the aforementioned taxonomic profiles using StrainPhlAn 4 (version 4.1)11,12. Those phylogenetic trees depict the relationship between conspecific strains detected in each sample of our cohort, with their phylogenetic distance being able to provide information on whether the same strain is observed in two samples. To reliably identify those strain-sharing events, we define SGB-specific strain boundaries. This is done by augmenting the phylogenetic trees with strains from publicly available longitudinal metagenomic datasets (n = 16 datasets including 4,322 stool metagenomes from 1,334 healthy individuals48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64) and by leveraging previous observations that, in close longitudinal samples (46. After this step, a strain-sharing matrix indicating how many strains are shared between each pair of samples in our cohort is generated. This is used to compute the DoSER, defined as the number of donor strains detected in the patients (considering all donor samples administered to the recipient up until that point) over the total number of SGBs typed at strain level in the donor. To allow comparison of the DoSER between arms, the main donor baseline aliquot administered to the d-FMT arm is used as reference for the p-FMT arm. The strain replacement rate was defined as the number of different strains profiled in longitudinal samples of the same individual over the common SGBs typed at strain level in both samples.
Randomization and masking
An online random number generator software (https://www.sealedenvelope.com) was used to provide random permuted blocks with a block size of four and an equal allocation ratio; the sequence was hidden until the interventions were assigned. Blocked randomization of patients was performed by an external individual not involved in the study.
To mask treatments to recipients, both infusate bottles and syringes were covered with dark-colored paper before the infusion, and the patients were sedated. Placebo capsules were made of cellulose and were identical in appearance to the FMT capsules, to ensure the blinding of patients and study staff. Moreover, both placebo suspension bottles and placebo capsule boxes were labeled with similar barcodes as FMT ones. Physicians who evaluated patients at follow-up (clinical outcome assessors) were blinded to administered treatments, as were authors involved in outcome analysis.
Sample size
Sample size calculation was based on the hypothesis of the superiority of FMT + standard of care (SOC) over SOC alone. The 1-year PFS rate for SOC has been reported to be nearly 60%. The alternative hypothesis is that FMT can improve the 1-year PFS rate from 60% to 80% when associated with SOC. A total of 50 patients is required to enter this two-treatment, parallel-design study. The probability is 80% that the study detects a treatment difference at a one-sided 5.0% significance level, if the true hazard ratio is 0.436. This is based on the assumption that the accrual period will be 36 months, the follow-up period will be 36 months and the median survival is 15.1 months. The total number of events is expected to be 36.
Statistical analysis
Data were summarized as median and range when referred to quantitative variables and as absolute counts and percentages if related to categorical items and reported according to the randomized arm. Survival curves were estimated with the Kaplan–Meier method and compared with the Breslow test. We chose this test as it gives more relevance to initial events. In our study, early disease progression would be more important than late disease progression, as the patient who experiences progression is not allowed to receive further FMTs, which are expected to strengthen the donor effect. Therefore, early events would be more important than late events in reducing the global efficacy of the donor FMT.
Median survival times were reported with their 95% CIs. A Cox proportional hazard model was implemented after checking the proportionality assumptions, and the hazard ratio with its 90% CI was presented. Differences between proportions were assessed with the chi-square test.
A non-parametric approach was followed to analyze microbiome features based on Wilcoxon signed-rank or Mann–Whitney test when applied to quantitative variables either paired or unpaired, respectively, and Fisherʼs exact test for categorical items. The PERMANOVA test evaluates whether the average microbiome structure or the compositional variability differs significantly between groups65. This test was applied on Bray–Curtis dissimilarity matrices to evaluate compositional microbiome differences between arms at baseline.
IBM-SPSS version 28.0 statistical software, GraphPad Prism version 10, R version 4.4.2 (‘survival’ and ‘survminer’ packages) and Python version 3.10.12 (‘scikit-bio’ and ‘scipy’ packages) were used for the analyses.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.