Citations

Background: Mosquito-borne viral diseases are a growing global health threat, and artificial intelligence (AI) and machine learning (ML) are increasingly proposed as forecasting tools to support early-warning and response. However, the available evidence is fragmented across pathogens, settings and modelling approaches. This review provides, to the best of our knowledge, the first comprehensive comparative assessment of AI/ML models forecasting mosquito-borne viral diseases in human populations, jointly synthesising predictive performance across model families and appraising both methodological quality and operational readiness. Methods: Following PRISMA 2020, we searched PubMed, Embase and Scopus up to August 2025. We included studies applying AI/ML or statistical models to predict arboviral incidence, outbreaks or temporal trends and reporting at least one quantitative performance metric. Given the substantial heterogeneity in outcomes, predictors and time–space scales, we conducted a descriptive synthesis. Risk of bias and applicability were evaluated using PROBAST. Results: Ninety-eight studies met the inclusion criteria, of which 91 focused on dengue. The forecasts spanned national to city-level settings and annual-to-weekly resolutions. Across classification tasks, tree-ensemble models showed the most consistent performance, with accuracies typically above 0.85, while classical ML and deep-learning models showed wider variability. For regression tasks, errors increased with temporal horizon and spatial aggregation: short-term, fine-scale forecasts (e.g., weekly city level) often achieved low absolute errors, whereas long-horizon national models frequently exhibited very large errors and unstable performance. PROBAST assessment indicated that most studies (63/98) were at high risk of bias, with only 24 judged at low risk and limited external validation. Conclusions: AI/ML models, especially tree-ensemble approaches, show strong potential for short-term, fine-scale forecasting, but their reliability drops substantially at broader spatial and temporal scales. Most remain research-stage, with limited external validation and minimal operational deployment. This review clarifies current capabilities and highlights three priorities for real-world use: standardised reporting, rigorous external validation, and context-specific calibration.

Abstract Background Watchful waiting is a common option for pts with asymptomatic FL/MZL. Anti-tumor immunization may delay or even avoid subsequent need for more toxic therapies. EO is designed from non-self-protein sequences derived from gut bacteria, including 4 HLA-A2 CD8 T cell epitopes (synthetic mimic peptides), exhibiting molecular mimicry with specific epitopes on B cell markers (CD20, CD22, CD37, BAFF-receptor). EO also contains a CD4 helper epitope UCP2. EO expands pre-existing memory CD8 T cells recognizing non-self-epitopes from gut bacteria that cross-react with B cell antigens on tumor cells. Methods Cohort 2 (C2; planned accrual 25 pts) of EONHL1-20/SIDNEY includes HLA-A2+ pts with previously untreated FL/MZL not in need of treatment. Pts receive EO (300μg/peptide) SC with adjuvant Montanide, q2 weeks (w) x 4, then q4w, for 12 total doses. For antigen-specific immune responses, peripheral blood mononuclear cells were assayed by flow cytometry using EO-mimic or B cell peptide specific tetramers without in vitro stimulation. IFN-γ ELISPOT was used to evaluate response to UCP2 and anti-CD3. Primary endpoint is objective response rate (ORR) per Lugano 2014. A pre-defined futility boundary was applied (ORR uninteresting=5%/promising=20%). Finding at least 3 of 25 pts with ORR would meet criteria to continue development (chance of 13% and 90% for true ORR of 5% and 20%, respectively). Results As of July 2025, 22 pts had started EO. Median age 59 years (range 32-86); 9 female; 19 ECOG 0/3 ECOG 1; 19 FL/3 extranodal MZL (ENMZL); 3 Ann Arbor stage I-II/8 stage III/11 stage IV; FLIPI low risk 8/intermediate 8/high 6; FLIPI-2 low risk 12/intermediate 7/high 3; 19 GELF neg/3 pos. At a median follow-up for progression of 9.4 months, 9 pts completed EO, 4 stopped for progressive disease (PD), 2 due to PI/pt decision; 7 ongoing. Median treatment duration 38 w (range 4-39). Among 19 evaluable pts, best Lugano response was 3 complete (CR) and 6 partial responses (PR) (ORR 47% [95% CI 24%–71%]; CR-rate 16% [3%-40%]), 6 stable disease (SD), and 4 PD. ORR by PET-CT was 29% (n=14) w6, 39% (n=18) w19, 46% (n=13) w42; corresponding CR rates were 0%, 11%, 15%. Median time to response 17.1 w (range 5.3-41.1). Currently there are only 2 confirmed PD events in pts with prior PR/CR. Among 8 pts with an initial indeterminate response per LyRIC 2016 followed by a confirmatory scan, 4 pts did not have confirmed PD; a possible atypical response pattern rate of 50% on EO (occurring 5.3-41.9 w after start of EO). In 16 immune response evaluable pts, 88% and 81% had expansion of CD8 T cells specific for EO mimic and B cell target peptides, respectively. Expanded specific CD8 T cells had a memory phenotype dominated by effector memory (TEM) and central memory cells. At w5, 7 clinical responders and 7 non-responders were evaluable; more responders had a positive immune response against EO mimic (Chi-square p=0.025) and B cell target (p=0.047) peptides. Max expansion (response sum for 4 EO mimic or 4 B cell target peptides at a single timepoint w3-w8) of specific CD8 T cells (% of all peripheral CD8 T cells) was higher for responders than non-responders both for EO mimic (Mann Whitney p=0.018) and B cell target (p=0.017) peptides. Pts (4) with an initial indeterminate response (LyRIC 2016) had a higher max expansion of B cell target peptide specific CD8 T cells vs 4 pts with a typical progression pattern (Mann Whitney p=0.018) and vs all other 12 pts evaluable for immune response (p=0.029). There was no difference in CD4 T cell response to UCP2 in responders vs non-responders (induction or max immunity w3-w8). Anti-CD3 expansion of T cells indicating T cell intrinsic activation potential did not correlate with clinical response (baseline or max). The most common EO-related adverse events (AEs) were Gr (Grade) 1 (10 pts) and Gr 2 (6 pts) local administration site reactions. Other related AEs in >1 pt were Gr 1 headache (3 pts) and Gr 1-2 fatigue (3 pts). Asthenia lasting 2 days was the only related Gr 3 AE. There were 4 unrelated Gr 3 AEs (gastroenteritis, GI-hemorrhage, inguinal hernia, and syncope). There were no Gr 4-5 AE. ConclusionsEO has a favorable safety profile and is preliminary associated with an ORR of 47%, exceeding the pre-specified boundary for promising activity. The speed and level of expansion of both EO mimic and B cell target peptide specific CD8 memory T cells are associated with objective response. Results will be updated.

Abstract Background Single-agent R is a common 1st-line therapy for patients (pts) with FL, especially those with low-tumor burden advanced stage disease or comorbidities. EO is a therapeutic vaccine generated from non-self-protein sequences from gut bacteria, including 4 HLA-A2 CD8 T cell epitopes that mimic B cell-specific markers: CD20, CD22, CD37 and BAFF-receptor. EO also contains a CD4 helper epitope UCP2. EO expands pre-existing memory CD8 T cells recognizing non-self-protein sequences from gut bacteria which can cross-react with B cell antigens on tumor cells. The aim adding EO to R is to safely increase the depth and duration of responses. Methods Cohort 3 (C3) of trial EONHL1-20/SIDNEY includes pts with HLA-A2 and previously untreated low-tumor burden (GELF) FL grade 1-3A in need of treatment (per pt/treating physician). Pts receive EO (300μg/peptide) SC with adjuvant Montanide, q2 weeks (w) x 4, then q4w for a total of 12 doses, combined with R starting at w7 (375 mg/m2 IV q1w x4, then q8w x4). For immune responses blood mononuclear cells were assayed by flow cytometry and EO-mimic or B cell peptide specific tetramers without in vitro stimulation. The primary objective is to assess safety; secondary objectives include EO immunogenicity and preliminary efficacy. Results As of July 2025, the 6 planned pts had started EO. Median age was 66 years (range 46-73); 5 ECOG 0/1 ECOG 1; 1 Ann Arbor stage III/5 stage IV; FLIPI low risk 2/intermediate 2/high 2; FLIPI-2 low risk 2/intermediate 4. At median follow-up 10.6 months (mo) 4 pts completed and 2 are ongoing on EO. Median treatment duration was 43 w (range 2-43). Best response by Lugano in the 6 pts included 4 complete (CR) and 2 partial responses (PR). Median time to PR/CR was 17.1 w (range 6.0-18.0). Currently there is only 1 (not yet confirmed) progression at 20.5 mo. All 5 immune response evaluable pts had expansion of CD8 T cells specific for EO mimic and B cell target peptides during treatment; 4 of 5 had detectable expansions at w5 (1st testing time). Currently, the longest tested immune response is at w65 and positive for both EO mimic and B cell target peptides. Expanded specific CD8 T cells had a memory phenotype predominantly composed of effector memory (TEM) cells but also including central memory (TCM) and terminally differentiated effector memory CD8 T (TEMRA) cells. Across the 5 pts, the medians of max % (sum of specific CD8 T cells targeting either the 4 EO mimic peptides or the 4 B cell antigen epitopes at a specific timepoint) among all peripheral CD8 T cells were 0.71% (range 0.19-3.97) for EO mimic and 0.33% (0.20-1.30) for B cell target peptides. The most common EO-related adverse events (AEs) were local administration site reactions (erythema, pain, induration). These included Gr (Grade) 1 reactions in 3 pts, Gr 2 in 1 pt, and Gr 3 in 1 pt; latter being ulceration with EO-interruption at w10.1, and in context of a strong immune response against EO mimic [3.97% of all peripheral CD8 T cells at w5] and B cell target [1.30%] peptides, and CR from w16.7. Other EO related AEs were Gr 1 lymph node pain (1 pt), and related to both EO and R were Gr 2 urticaria (1 pt) and Gr 1 fatigue (1 pt). Only R related were Gr 2 infusion related reaction (3 pts), Gr 2 enterocolitis infection (1 pt), and in 1 pt each Gr 1 flushing, anemia, headache and diarrhea. There were no further Gr 3-4 AE and no death on treatment/follow-up. Complete B cell depletion was seen in 5 pts tested at w18 (R added w7); no recovery in the range of w30-65. Only 3 infections were reported, per above enterocolitis, and 2 cases of COVID-19 both resolved without sequalae. ConclusionsThe combination of EO with R has a predictable and manageable safety profile, with EO only adding local administration site reactions to the well-known R safety profile. In this feasibility cohort all evaluable pts developed a specific immune response against EO and B cell targets, and a Lugano objective response. The combination of EO and R is feasible and can be evaluated in further trials. Updated results will be presented at the meeting.

Abstract Objective To examine the influence of demographics on organ damage in a broad, multiethnic cohort of patients with Behçet’s syndrome (BS). Methods In this cross-sectional ancillary analysis of the PROBE project, the investigated demographic variables were sex, age, education level and geographic area of residence. Damage was measured by the BS Overall Damage Index (BODI). Multivariate linear (β) and logistic (adjusted odds ratio [adjOR]) regression analyses examined associations between demographics and the extent and prevalence of damage. Results A total of 970 patients were enrolled. The median (interquartile range) age was 40 (31–50) years; 56.5% were males; 21.4% had a low level of education. The median BODI score was 1 (0–3), with 65.6% of patients having a BODI ≥1. Males had higher damage (β 0.103) and a higher prevalence of total (adjOR 1.7, per 10 years), ocular (adjOR 1.6), and vascular (adjOR 2.1) damage. Age was associated with greater damage (β 0.104), and a higher prevalence of overall (adjOR 1.4), neuropsychiatric (adjOR 1.2) and miscellaneous (adjOR 2.0) damage. Low education was associated with a greater frequency of overall (adjOR 1.9) and ocular (adjOR 1.6) damage. North African patients experienced greater damage than South European (β 0.400) and Middle Eastern (β 0.314) patients, as well as a higher risk of overall damage (adjOR 13.9 and 10.7, respectively) across all organ domains, except for the reproductive and gastrointestinal systems. Conclusions This study demonstrates how the extent, prevalence and characteristics of damage in BS vary with demographic factors, underscoring their importance in research and personalized management.

Abstract Outcomes in classic Hodgkin lymphoma (cHL) have steadily improved; however, additional therapies are needed for patients who relapse or do not respond to novel agents. Here, we report the efficacy and safety of camidanlumab tesirine (Cami), an anti-CD25 antibody-drug conjugate, in patients with relapsed/refractory cHL after brentuximab vedotin/programmed cell death protein 1 inhibitor therapies from the phase 2 ADCT-301-201 study. Eligible patients were adults with cHL who had received ≥3 previous lines of systemic therapy (or ≥2 if ineligible for hematopoietic stem cell transplant). Patients received 45 μg/kg Cami (IV, once every 3 weeks) in cycles 1 to 2, followed by 30 μg/kg IV once every 3 weeks for ≤1 year. The primary end point was overall response rate (ORR) per 2014 Lugano classification. Secondary end points included complete response rate (CRR), progression-free survival (PFS), and overall survival (OS). In total, 117 patients were enrolled with a median age of 37.0 years (range, 19-87). The ORR was 70.1% (95% confidence interval [CI], 60.9-78.2), with a CRR of 33.3% (95% CI, 24.9-42.6). The median PFS was 9.13 months (95% CI, 5.3-15.0); median OS was not reached. Thirty-three (28.2%) patients discontinued treatment because of treatment-emergent adverse events; the most common reasons were skin and subcutaneous tissue disorders (n = 10 [8.5%] patients), infections and infestations (n = 5 [4.3%]), and nervous systems disorders (n = 5 [4.3%]). Guillain-Barré–type or polyradiculopathy-type events occurred in 8 (6.8%) patients. Cami was efficacious in this heavily pretreated population; however, the efficacy was overshadowed by substantial issues with the safety profile. This trial was registered at www.clinicaltrials.gov as #NCT04052997.

Abstract Background Tafamidis is a kinetic stabilizer that binds to the transthyretin (TTR) gene, inhibiting its dissociation. It is the only disease-modifying treatment for hereditary TTR amyloidosis with peripheral neuropathy (ATTRv-PN) available in the National Therapeutic Form (Formulário Terapêutico Nacional, FTN, in Portuguese) of the Brazilian Unified Health System (Sistema Único de Saúde, SUS, in Portuguese). Objective To assess if the efficacy and safety of tafamidis in the Brazilian real-world experience are comparable to the results of clinical trials. Methods We retrospectively studied all patients with ATTRv-PN evaluated at our center from September 2011 to March 2022 (data cut-off) who were initiated on tafamidis and had at least 1 follow up visit 6 months after the initiation of the drug treatment. Neurologic and functional outcomes were compared from day 1 (D1) of the tafamidis treatment (baseline) to the last follow-up. Results In total, 33 patients were included, 18 (55%) of whom were female. All patients were carriers of the V30M mutation of ATTRv-PN, and 20 (61%) presented early onset (EO) ATTRv-PN. At baseline, the median age of the sample was of 40 (interquartile range [IQR]: 36–68) years, the median Neuropathy Impairment Score (NIS) was of 10 (6–24) points, and the median body mass index (BMI) was of 26 (23–28) kg/m2. The mean follow-up time was of 3.4 years. At the last follow-up, the BMI, the neurological impairment, and the level of disability slightly worsened compared with baseline, while the findings of the nerve conduction studies remained stable. These same results were observed across EO and late-onset (LO) ATTRv-PN patients. A total of 25 (75.8%) patients were considered responders, and 8 (24.2%), non-responders. Conclusion The efficacy and safety of tafamidis reported in clinical trials is expandable to the Brazilian real-world scenario in EO and late-onset (LO) ATTRv-PN.

Background Follicular lymphoma (FL) is a generally indolent disease where a watch and wait strategy is standard in asymptomatic patients (pts) with a low tumor-burden. Harnessing antitumor immunity through an anti-tumor immunization strategy at this stage may delay or avoid the subsequent need for more toxic and complex therapies. EO2463 is a therapeutic vaccine designed from non-self protein sequences, derived from gut bacteria, including 4 HLA-A2 synthetically produced 9-mer CD8 cytotoxic T cell epitopes, exhibiting molecular mimicry with specific epitopes on the B cell markers CD20, CD22, CD37, and CD268 (BAFF-receptor). Additionally, EO2463 contains a CD4 helper epitope UCP2 derived from hTERT. The compound expands pre-existing memory CD8 T cells recognizing the non-self protein sequences from gut bacteria which can then cross-react with the selected B cell antigens on tumor cells. EO2463 is being used to drive anti-tumor immunity against B cell malignancies, and has demonstrated long-lasting, specific CD8-memory T cell responses and clinical activity in relapsed/refractory FL and marginal zone lymphoma (MZL) [J Clinical Oncol 2024, 42 (S16), 7058]. Furthermore, CD8 T cells expanded from samples of patients treated with EO2463 can kill HLA-A2 expressing lymphoblastoid cell lines presenting the B cell targets [Hematological Oncol 2023, 41 (S2), 581-582]. Methods The current report describes Cohort 2 of trial EONHL1-20/SIDNEY (NCT04669171) and includes pts with newly diagnosed, previously untreated FL, grade 1-3A, HLA-A2, and not in need of treatment. Pts receive EO2463 (300μg/peptide) SC with adjuvant Montanide ISA 51 VG, q2 weeks (w) x 4, then q4w, for a total of 12 doses (39w; PET-CTs at w6, w19, and w42). Planned accrual for this cohort is 25 pts. Immune responses were assayed on thawed pt peripheral blood mononuclear cells by flow cytometry and EO2463 peptide specific tetramers or by IFN-γ ELISPOT (assaying responses against 9-mer peptides from CD20, CD22, CD37, and CD268; and from peptides of EO2463) without any proceeding in vitro stimulation (ex vivo). Results As of June 2024, 15 pts had started study treatment. Median age was 62 (range 55-66) year, 7 male and 6 female, 2 pts Ann Arbor stage I/II, 5 stage III, and 8 stage IV. Ten of 15 pts were still on treatment; median treatment duration was 18.4 weeks (range 5-39 w) for all pts. Among 13 evaluable pts, best responses by PET-CT using Lugano criteria were 1 pt CR, 5 pts PR (46% objective responses (OR)), 3 pts SD, and 4 pts PD. Of those with PD per Lugano, 2 had isolated increases in FDG-uptake without increase in lesion size/number at w6 and both remain on treatment (per protocol based on LYRIC criteria). OR was first noted w6 in 3 pts and w19 in 3 pts; duration of response is not evaluable at this time due to short follow-up. The pt with CR had isolated increases in FDG-uptake at w6, PR at w19, and CR at w35. Anti-EO2463 immune responses were observed in 4 of 5 tested pts at the 1st testing timepoint (w5) and onwards (currently longest tested up until w31). In an exploratory analysis, the percentage of EO2463 peptide specific CD8 T cells measured by tetramers seemed linked to clinical response (OR 3 pts vs non-OR 2 pts; 2-sided T-test, p=0.02; range of percentages in pts with OR 0.85%-1.39% of all CD8 T cells in peripheral blood, range in pts with non-OR 0.17%-0.27%). Functionality and cross-reactivity of responses were shown by IFN-γ ELISPOT using 9-mer peptides corresponding to the B cell targets in 4 tested pts. Expanded antigen-specific CD8 T cells displayed a memory phenotype; effector/memory (TEM), and terminally differentiated effector memory CD8 T cells (TEMRA). The most common related AEs were grade 1 (7 pts) and grade 2 (5 pts) local administration site reactions (mainly erythema, pruritus, and induration). Other related AEs in > 1 pt were grade 1 headache (2 pts) and grade 1-3 fatigue/asthenia (3 pts). Asthenia lasting 2 days was the only related grade 3 adverse event (AE). Conclusions EO2463 monotherapy is associated with a very manageable safety profile and appears to demonstrate meaningful single-agent efficacy in FL. Consistent with the preclinical hypothesis, expansion of specific CD8 memory T cells may be associated with response. Results will be expanded at the meeting. Study EONHL1-20/SIDNEY also includes cohorts, not covered in this abstract, exploring EO2463 plus rituximab for 1st line FL/MZL, and EO2463 plus lenalidomide/rituximab in relapsed FL/MZL.

Background Follicular and marginal zone B cell lymphoma (FL; MZL) demonstrate an indolent course with heterogeneous outcomes and a potential for spontaneous remissions indicating immune system intervention. T cell-based therapeutic approaches are therefore an attractive proposition in FL and MZL. EO2463 peptide immunotherapy expands pre-existing memory CD8 T cells recognizing non-self-protein sequences from gut bacteria which cross-react with B cell antigens. The compound includes 4 HLA-A2 synthetically produced epitopes which exhibit molecular mimicry with the B cell markers CD20, CD22, CD37, and CD268 (BAFF-receptor), as well as a the CD4 helper-epitope UCP2 derived from hTERT. EO2463 is being used to drive anti-tumor immunity against B cell malignancies, and has demonstrated long-lasting, specific CD8-memory T cell responses and clinical activity in relapsed/refractory FL and MZL [J Clinical Oncol 2024, 42 (S16), 7058]. Furthermore, CD8 T cells expanded from samples of patients (pts) treated with EO2463 can kill HLA-A2 expressing lymphoblastoid cell lines presenting the B cell targets [Hematological Oncol 2023, 41 (S2), 581-582]. Methods The current report describes testing of a possible biomarker in the phase 1 portion of the ongoing EONHL1-20/SIDNEY trial (NCT04669171).This part of the trial (“Cohort 1”) included pts (n=9) with relapsed (at least 1, no limit, prior treatments) FL and MZL, stage 1-3A, confirmed to be positive for HLA-A2. Pts received EO2463 (150 [n=3] or 300μg/peptide [n=6]) SC with adjuvant Montanide ISA 51 VG q2 weeks (w) x 4, then q4w, for a max of 12 months; at w7 lenalidomide (20 mg/day for 21/28 days up to 12 cycles) was added (EL), and if no complete remission (CR) at w19, rituximab (375 mg/m IV, q1w x 4, then q4w x 4) was also added (ER2). Immune responses and clinical outcomes of Cohort 1 were previously reported [J Clinical Oncol 2024, 42 (S16), 7058]. To assess a possible biomarker, tetramers were used to specifically detect CD8 T cells recognizing the 4 epitopes in the EO2463 compound (derived from gut bacterial proteins exhibiting molecular mimicry with the B cell markers CD20, CD22, CD37, and CD268). The tetramers were all marked with phycoerythrin. Thawed samples of peripheral blood mononuclear cells taken at start of w5 (after 2 administrations of EO2463 monotherapy, each followed by 2w observation) were stained ex vivo (without any in vitro stimulation) with the tetramers and assayed by flow cytometry. Results are presented as percentage of EO2463 specific CD8 T cells among all CD8 T cells in the peripheral blood of the pt. Results Among the 9 treated pts, PET-CT at w6 (after 3 administrations of EO2463 monotherapy) suggested possible clinical activity in 4/9 pts (1 PR, 1 pt size reduction 15%, 1 pt 20% reduced tracer uptake, 1 pt 5/6 target lesions reduced metabolic activity). By PET-CT at w42 (after 12 administrations of EO2463 monotherapy, 11 cycles of lenalidomide, and rituximab treatment) the Lugano response assessment was complete response in 6/9 pts. A positive biomarker response was assessed in 5/9 pts (percentage of specific CD8 T cells ranged from 0.2-2.1%). EO2463 monotherapy PET-CT activity at w6 (n=4/9 pts): 4 of 4 pts with clinical activity were biomarker positive and 1/5 of the other pts was also positive; this pt had an isolated increase in FDG-uptake at w6, followed by a CR at w19. Four further other pts were biomarker negative. In this limited sample, the biomarker had sensitivity 100% and specificity 80%. EO2463 in combination with lenalidomide/rituximab PET-CT CR at w42 (n=6/9): 5 of 6 pts with CR were biomarker positive and 1 pt with CR was negative; the pt with a negative test had no objective tumor response until after rituximab. In this limited sample, the biomarker showed sensitivity 83% and specificity 100%. Conclusions The current preliminary assessment indicates that direct ex vivo tetramer staining for specific CD8 T cells after short term (5 w) treatment with EO2463 monotherapy could be a possible biomarker for clinical response, both for EO2463 monotherapy response, and for response to EO2463 in combination with lenalidomide/rituximab. We plan to validate these findings in the ongoing study cohorts of the trial; expansion phase 2 cohorts, exploring EO2463 monotherapy in the “watch & wait” setting, EO2463 plus rituximab for 1st line FL/MZL, and EO2463/lenalidomide +/- rituximab in relapsed FL/MZL.