Pathologic Responses to Stereotactic Ablative Radiotherapy in Combination With Nivolumab for Early Stage NSCLC: A Phase 2 Study.

Introduction
Stage I NSCLC accounts for approximately 20% cases of lung cancer, but screening advancements are shifting diagnosis to earlier stages.1,2 Despite improved prognosis for stage IA1 (92% 5-y survival), stage IB survival drops to 68%, highlighting the need for better regional and distant control.3 Surgery remains the preferred treatment for patients with adequate cardiopulmonary function.4 However, many are ineligible due to comorbidities, making stereotactic ablative radiotherapy (SABR) the standard alternative.5,6 Although SABR ensures more than 90% local control for small tumors, larger tumors pose a higher risk of regional and distant relapse.7, 8, 9
Neoadjuvant SABR studies for stage I NSCLC have revealed only a 60% pathologic complete response (pCR), well below the expected 90%.10 Meanwhile, immunotherapy with immune checkpoint inhibitors (ICIs) has transformed NSCLC treatment in stages II to IV.11 Combining radiotherapy and immunotherapy may enhance tumor control by disrupting the microenvironment, promoting tumor cell destruction, and triggering systemic immune activation.12,13 Early trials combining PD-1 inhibition with SABR demonstrated promise, leading to randomized phase 3 trials for patients with medically inoperable NSCLC.14 Most of these trials, however, enroll patients in an all-comer fashion, without considering high-risk features or biological characteristics that could influence immune resistance or a proinflammatory tumor microenvironment. In addition, by focusing solely on nonsurgical candidates, these trials inherently select patients with poorer overall fitness, which can mask long-term overall survival (OS) benefit of a more intensive approach. Finally, it remains unclear what microenvironmental changes and pathologic responses these combinations elicit, whether a more intensified approach is warranted for surgical candidates with higher risk stage I NSCLC, or whether surgery could be safely omitted in operable patients who demonstrate a robust response to a noninvasive strategy.
To address this gap, we conducted a phase 2 trial in stage I NSCLC, evaluating SABR plus three doses of neoadjuvant nivolumab, followed by surgery. Our primary objective was to determine the pCR rate compared with SABR alone. Furthermore, we aimed to assess microenvironmental changes and long-term outcomes while identifying biomarkers predictive of treatment response and resistance.

Methods

Study Design
We conducted a phase 2, open-label, single-arm study to evaluate whether neoadjuvant nivolumab combined with SABR improves pCR rates in patients with early stage NSCLC (tumors ≤4 cm). The trial took place from November 2019 to February 2022 at Hospital Israelita Albert Einstein (HIAE) and Hospital Municipal Vila Santa Catarina in São Paulo, Brazil. Approved by the HIAE Ethics Board, all participants provided written informed consent. The trial was registered at clinicaltrials.gov (NCT04271384).

Patients
Eligible patients were above or equal to 18 years of age with stage I NSCLC (≤4 cm, ≥1 cm solid component, cT1–T2a cN0 by American Joint Committee on Cancer eighth edition). Positron emission tomography–computed tomography (PET-CT) and invasive staging, when indicated, confirmed the absence of lymph node involvement. Lesions had to be suitable for SABR, as assessed by a radiation oncologist. Patients required adequate cardiopulmonary function for surgery. Exclusions included contraindications to surgery, active immunosuppressive/autoimmune disease requiring treatment, systemic corticosteroids more than 10 mg per day prednisone equivalent, or biological agents in the past 2 years.

Procedures
Nivolumab (360 mg) was administered every 21 days for three doses or until unacceptable toxicity. SABR began with the first nivolumab dose, lasting 2 to 3 weeks per a risk-adapted protocol (Supplementary Table 1).15 Safety assessments occurred every 21 days during treatment and 21 to 28 days post-treatment. PET-CT was repeated before surgery (10+-2 wk post-SABR). Surgery included video-assisted or robotic-assisted lobectomy or anatomical segmentectomy plus lymph node dissection.

Outcomes and Assessments
The primary end point was pCR at surgery. Secondary end points included major pathologic response (MPR) rate, safety (Common Terminology Criteria for Adverse Events version 4.0), event-free survival (EFS), OS at 12 months, and surgical outcomes. Because Response Evaluation Criteria in Solid Tumors criteria do not apply to irradiated lesions, objective response rate was not formally assessed. Safety was monitored until 100 days post-nivolumab, with 6-month chest CT follow-ups. Additional end points included immune profiling via peripheral blood flow cytometry, molecular changes related to pCR, and treatment resistance mechanisms. Plasma was collected at four time points for circulating tumor DNA analysis: baseline, before surgery, 4 weeks after surgery, and at 12 months.

Pathologic and Survival Assessments
pCR was defined as no viable invasive carcinoma in the surgical specimen. MPR was less than 10% viable tumor cells, a predictor of long-term NSCLC outcomes.16 Post-radiotherapy response evaluation relied on hematoxylin-eosin staining and microscopy. EFS was measured from nivolumab initiation to radiologic/clinical progression or death, and OS from treatment start to death. Patients alive and EFS at the last follow-up (minimum 12 mo) were censored.

Flow Cytometry
Peripheral blood tube samples with an ethylenediaminetetraacetic acid anticoagulant were obtained from 22 patients pre- and post-nivolumab treatment, and these samples were used to determine the percentage and absolute numbers of circulating T regulatory cells and T cells expressing PD-1, CD69, or both.
Briefly, fresh blood (100 μL) was lysed, washed, and stained using BD Pharmingen/Horizon antibodies: CD45-Alexa 700 (clone: HI30), CD3-PerCP-Cy5·5 (clone: SK7), CD4-APC-Cy7 (clone: SK3), CD8-PE-Cy7 (clone: RPA-T8), CD25-PE-CF594 (clone: M-A251), CD127-FITC (clone: HIL-7R-M21), CD69-BV-650 (clone: FN50), CD279/PD-1-BV711 (clone: EH12.1), and intracellular FoxP3-PE (clone: 259D/C7), all from BD Pharmingen, BD Horizon, or BD Biosciences. Intracellular FoxP3 staining followed fixation and permeabilization with BD Pharmingen’s human foxP3 buffer set.

Targeted Sequencing for FFPE Tissue and cfDNA
A custom panel targeted frequent NSCLC genomic alterations (ALK, BRAF, CDKN2A, EGFR, ERBB2, KRAS, LRP1B, MET, NOTCH1, PIK3CA, TP53). Formalin-fixed, paraffin-embedded (FFPE) samples underwent amplicon sequencing (150 bp amplicons), and cell-free DNA (cfDNA) libraries were prepared using Ion AmpliSeq Library Kit v.2.0 and Oncomine Pan-Cancer Cell-Free Assay.

Assessment of Gut Microbiota Using 16S rRNA Sequencing
Fecal samples (pre- and post-treatment) were stored at −80°C. DNA extraction used ZymoBIOMICS DNA kits, and 16S rRNA sequencing (V3V4 regions) followed Illumina protocols.17 Bioinformatics analysis was performed on Galaxy (usegalaxy.org) with Mothur v.1.39.5.18 Taxonomic classification used Silva 138.1.19 Relative abundances of Bifidobacterium, Akkermansia, and Ruminococcaceae—associated with PD-1 inhibitor efficacy—were compared between pCR and non-pCR patients.20

Statistical Analysis
Sample size followed Simon’s 2-stage design, assuming a 60% historical pCR rate with SABR alone and a 40% improvement with nivolumab-SABR (targeting 84% pCR). With a 5% alpha (two-sided) and 80% power, 28 patients were needed. Accounting for 10% dropout, 30 were to be enrolled, but the trial terminated early due to coronavirus disease 2019 (COVID-19)–related slow accrual.
Formal statistical testing was limited to the primary end point. pCR rates were tabulated with 95% confidence intervals (CIs). EFS and OS were estimated using Kaplan-Meier curves. Cox regression analyzed baseline characteristics' correlation with pCR. Wilcoxon tests compared immune cell infiltration, microbiome changes, T regulatory cells, and PD-1/CD69 T-cell expression pre- and post-treatment. All CIs, except for the primary end point, were two-sided and set at 95%.

Results
A total of 25 participants were enrolled in the study from November 2019 to February 2022 (Fig. 1). However, 24 successfully completed the experimental therapy and subsequently underwent surgery. One patient did not proceed to surgery after SABR with nivolumab, due to death from relapsed acute alcoholic hepatitis after the first dose, unrelated to the study treatment. The median follow-up at the data cutoff was 12.1 months.

Patient Characteristics and Treatment
Baseline characteristics for all 25 enrolled patients are summarized in Table 1. The median age was 66 years. Most patients were female (68%), and 88% were current/former smokers. Most tumors were adenocarcinomas (N = 18 [72%]), and the median largest diameter of tumor was 2.30 cm (2.00; 3.00). The SABR doses delivered included 54 Gy in three fractions (N = 4 [16%]) and 50 Gy in five fractions (N = 21 [84%]). No patients had central/ultracentral tumors. All three doses of nivolumab were completed by 24 (96%) of the 25 patients. Most patients (N = 18 [75%]) underwent video-assisted thoracoscopic surgery (VATS) and six [25%] underwent a robotic-assisted approach. Two patients were converted from VATS to an open approach intraoperatively, one of whom required a reoperation due to complications. Both patients who required conversion died from surgical complications. There were no significant treatment delays (mean time to surgery from the last dose of 81.6 d, Supplementary Table 2).

Pathologic Assessment and Response
pCR was achieved in 19 of 24 cases that underwent surgery (79.2%; 95% CI: 57%–92%, p value = 0.0875, in the per-protocol analysis, Fig. 2). MPR was observed in 20 cases (83.3%; 95% CI: 61.8%–94.4%). Four patients did not achieve an MPR, one of whom had 100% of residual viable tumor (no pathologic regression). The tumor from this patient did not demonstrate any anatomical or metabolic improvement in FDG uptake by PET-CT but was fully resected. As aforementioned, no formal radiologic response was assessed, as the single target lesion was submitted to ablative radiotherapy. However, all other patients had either radiologic improvement or inflammatory features impairing proper tumor evaluation, which largely underestimated pathologic response.

Survival Outcomes
The median EFS and OS were not reached. At the time of data cutoff, there were no relapses, with a locoregional and distant tumor control rate of 100%. EFS and OS at 12 months were 84% (Fig. 3A and B). Six patients died during the study and follow-up period, none from disease progression.

Adverse Effects
All treatment-emergent and treatment-related adverse events are listed in Table 2. There were 29 adverse events, most of which were grade 1 (n = 20; 69%) or 2 (n = 4; 13.8%), being 16 possibly or probably related to the study therapy. No cases of grade 4 or 5 adverse effects occurred related to nivolumab or radiotherapy. There were no cases of immunotherapy- or radiation-related pneumonitis. Two patients presented with grade 1 cough and dyspnea, possibly attributed to nivolumab or radiation, but without radiologic findings to classify them as pneumonitis. Two patients (8%) had hypothyroidism (grades 1 and 2), one of whom had transient hyperthyroidism. There were no major adverse events from SABR. Two treatment-related grade 3 diarrhea events occurred, related to nivolumab.
Regarding treatment-emergent deaths and surgical outcomes, one patient died from relapsed alcoholic hepatitis after the first dose of nivolumab and was the only patient who did not undergo surgery, as mentioned. During the adverse event, this patient underwent a liver biopsy that revealed no signs of immune-mediated liver damage and confirmed alcoholic hepatitis-related features. Given the compatible pathologic findings, prior history of alcoholism, and acute alcohol intoxication before presenting to the emergency department, this grade 5 event was deemed unrelated to study therapy.
In addition, two patients died from surgical complications. One patient sustained a pulmonary artery injury during VATS, requiring conversion to open surgery and a subsequent reoperation for thoracic bleeding. Despite prolonged hospitalization, the patient ultimately died from complications beyond the 100-day safety follow-up window. After thorough video review by the surgical team, there was mild peribronchial inflammation that could have hindered arterial dissection and may have been caused by preoperative immunotherapy, though the patient was a current heavy smoker. A second case required conversion to open surgery, also during VATS, caused by an injury of an anomalous artery. The patient’s bleeding was controlled, and no further complications were observed during follow-up. The second grade 5 event was a fatal pulmonary embolism that occurred within 30 days postoperatively, after hospital discharge, and an uneventful immediate postoperative recovery. This event may have been associated with a prior COVID-19 infection, which occurred before the patient’s initial dose of nivolumab.

Biomarker Analysis
Eight patients had sufficient tissue for sequencing, and 20 patients were submitted to cfDNA testing. Only two patients had detectable mutations in cfDNA at baseline, one of them carrying pathogenic variants in SMAD4 and GNAS, and the second patient had a PIK3CA-activating mutation. Both patients had complete pathologic regression (Supplementary Table 4). Of note, cfDNA analysis is less sensitive in early stages of cancer. In fact, none of the samples from other time points after neoadjuvant treatment or surgery had detectable circulating tumor DNA. Of the eight patients with enough cancer tissue for sequencing, only one had no evidence of pathologic regression (with 100% viable tumor cells). For this patient, an EGFR exon 19 deletion was found in the tissue sample (Supplementary Table 4).
In this small sample size, no specific pattern of molecular alterations was found. Two patients presented more than one pathogenic variant in tissue, one of them harbored a TP53 loss-of-function mutation and PIK3CA-activating mutation, and the other one in KRAS (G12C), CDKN2A loss of function and TP53 loss of function. For the remaining patients, harboring one different mutation each, one patient had a mutation in EGFR, one patient had a MET exon 14 skipping mutation, one patient had an activating mutation in PIK3CA, and one patient had a loss-of-function mutation in TP53.
No significant changes were observed in the overall fecal bacterial composition (β-diversity) between complete and incomplete pathologic responses. We searched specifically for Bifidobacterium, Akkermansia, and Ruminococcaceae, known to enhance PD-1 treatment in NSCLC.20, 21, 22 We observed that the genera Bifidobacterium, Akkermansia, and unclassified Ruminococcaceae were more associated with pCR when compared with pathologic incomplete response (Fig. 4A and B). Akkermansia was found more abundant in patients with complete pathologic response, but none of the patients with incomplete pathologic response presented any levels of Akkermansia in the gut. Nevertheless, due to the small number of patients, no statistical significance could be achieved between complete and incomplete pathologic responses.
Interestingly, two patients with different outcomes—the one with no evidence of any pathologic regression and other with a complete pathologic response—shared the same EGFR mutation. Notably, these two patients had a marked difference in the relative abundance of Bifidobacterium, Akkermansia, and unclassified Ruminococcaceae in their stools. The patient with no evidence of pathologic regression presented a reduction in the relative abundance of unclassified Ruminococcaceae and a depletion of the genera Bifidobacterium and Akkermansia in stool samples.
Eight patients were assessable for PD-L1 expression. Two (25%) were PD-L1 negative, four (50%) had a low expression (1%–49%), and two (25%) had PD-L1 greater than 50%.
Peripheral blood flow cytometry demonstrated statistically significant T-cell PD-1 depletion in post-treatment samples (0.9% [0.1%–4%]) versus pretreatment (9.7% [1%–28%]), as well as an increase on T cells expressing CD69, from 10.4% (1.9%–26.3%) to 18% (3%–58.9%), inferring T-cell activation (Fig. 5A). In addition, T-regulatory cells also decreased from 1.9% (0.42%–10.2%) to 0.8% (0.1%–4.3%) with immunotherapy treatment in all but the single patient with absence of pathologic regression (Fig. 5B).

Discussion
In this trial, we investigated the efficacy of neoadjuvant nivolumab in conjunction with SABR followed by surgery in patients diagnosed with early stage NSCLC. Despite the study's limitation in sample size, which affects statistical robustness, our findings revealed a noteworthy 79% pCR rate when surgery was performed 10 weeks post-treatment completion. We achieved a 100% locoregional and distant control rate at 12 months. However, two patients died from surgical complications. Our results indicate that the neoadjuvant regimen is highly effective, highlighting the emerging role of adding immunotherapy to SABR for patients who are not candidates for surgery and raising the question of whether this treatment would suffice to eradicate early stage tumors while sparring a patient from surgery.
Immunotherapy has significantly transformed the landscape of outcomes in NSCLC. Recent adjuvant trials targeting stages IB to IIIA disease (American Joint Committee on Cancer seventh edition) have revealed improvements in recurrence-free survival rates. However, the modest absolute benefit, absence of a significant OS advantage, prolonged treatment duration, and the emergence of immune-related adverse events during extended adjuvant therapy have tempered enthusiasm in clinical practice.23,24 However, the neoadjuvant approach has garnered renewed attention in the era of immunotherapies.15 The preserved vascular capillarity and the presence of the tumor (and its neoantigens) may enhance preoperative immunotherapy activity when compared with a similar approach, but in the adjuvant setting.25 In NSCLC clinical studies, crosstrial comparisons between phase 3 trials have revealed a more substantial improvement versus placebo with neoadjuvant/perioperative immunotherapy, both in terms of EFS and OS, when compared with adjuvant immunotherapy trials, an impression confirmed by meta-analyses.26, 27, 28, 29
Another concept that has resurfaced in NSCLC is the evaluation of pathologic response. Historically, although pCR rates correlated with survival, they remained below 5% with chemotherapy, therefore not often pursued by clinicians.30 However, contemporary studies have unveiled more promising prospects, with pCR rates ranging from 10% to 20% with two to four doses of anti–PD-(L)1 alone and escalating to 20% to 60% with chemoimmunotherapy. This observed increase in pCR rates aligns with favorable long-term outcomes in patients with NSCLC.26, 27, 28,31,32
The integration of radiotherapy with chemotherapy has demonstrated improved pCR rates in locally advanced NSCLC.33 However, owing to its localized effect, radiotherapy has demonstrated no impact on distant recurrences or OS.34 Palma et al.14 demonstrated a 60% pCR rate when SABR alone was administered before surgery. Interestingly, despite the authors' estimation of a true pCR rate of 90% based on an excellent historical control with this treatment in nonsurgical candidates, the actual results fell short. Although a plausible explanation could be that 10 weeks may not be sufficient to assess the complete radiotherapy effect, which may take several months for maximal response, the observed suboptimal regional (53%) and distant (76%) control rates suggest that micrometastatic disease still represents the main cause for concern in early stage lung cancer.
Regarding the combination of stereotactic radiotherapy and immunotherapy, two trials yielded noteworthy results. First, durvalumab was evaluated in a neoadjuvant setting in a randomized phase 2 trial, administered either alone or in combination with SBRT at a non-ablative dose of 3 fractions of 8 Gy, followed by surgery.35 The combination demonstrated a pCR rate of 50%, in contrast to none in the durvalumab alone group. The lower pCR rate observed in this study, compared with our protocol, may be attributed to the lower radiotherapy dose, likely chosen due to the potential synergism of SBRT at non-ablative doses and immunotherapy, as previously described.36, 37, 38 We decided to use standard ablative doses to evaluate the full impact of a treatment frequently used for patients who are not candidates for curative surgery.
The second trial worth discussing is the I-SABR trial, a randomized phase II study led by MD Anderson Cancer Center, where SABR was administered concomitantly with four doses of nivolumab for patients ineligible for surgery.13 The authors reported a significant hazard ratio of 0.38 favoring the addition of nivolumab, with an absolute gain of 24% in EFS at 4 years (53%–77%). The improvement observed in this study mirrors the increase in our pCR rate from historical control (60%–79%), potentially reflecting a true benefit of adding nivolumab not only to local control but also to regional and distant control.
Although our trial was designed to assess operable patients, with the primary objective of evaluating the pCR rate after a neoadjuvant approach, our findings may inform broader discussions on how to incorporate this combination strategy in stage I NSCLC. Notably, our high pCR rates and favorable regional and distant control raise the possibility that adding immunotherapy could enable a less invasive treatment pathway—particularly relevant given the two postoperative deaths observed in patients who had already achieved a pCR, raising the question of whether a trimodality therapy is not necessary for most patients. Immunotherapy can complicate surgical resection in locally advanced tumors, though this was not observed in most of our patients, likely because all were N0. In two cases, peribronchial inflammation appeared to hinder arterial dissection, resulting in intraoperative injury and conversion to open surgery, one of whom died from further clinical complications. Whether this inflammation was related to prior immunotherapy or to the patients’ active smoking status remains uncertain. Notably, both events occurred during VATS, suggesting that robotic-assisted surgery may offer a safer approach for patients undergoing resection after immunotherapy.
Several phase 3 trials evaluating the addition of one year of immunotherapy to SABR have completed accrual in patients with medically inoperable NSCLC. However, two such studies have recently reported negative results. The KEYNOTE-867 trial, which assessed the addition of pembrolizumab for 1 year post-SABR, revealed no improvement in EFS and a numerically worse OS compared with placebo. Subgroup analyses indicated minimal benefit among patients above 65 years, those treated in East Asia, and never-smokers. Similarly, the SWOG/NRG S1914 trial demonstrated no advantage in integrating atezolizumab, despite a strategy involving priming 6 weeks before SABR, concurrent administration, and maintenance for a total of 6 months. Both studies also reported increased toxicity, including 1% to 2% grade 5 immunotherapy-related events.
Although our trial also enrolled patients in an all-comer fashion, most were current heavy smokers and surgical candidates. Our findings suggest that a more refined selection strategy—targeting patients with high-risk features and avoiding immunotherapy in tumors with immune-resistant biology, such as those harboring oncogenic drivers or arising in never-smokers—may yield better outcomes. Furthermore, extending immunotherapy duration may increase toxicity and costs without additional benefit, as indicated by the excellent efficacy outcomes from a short-course treatment in our data and supported by the I-SABR trial.
The full biomarker analysis is currently underway, but we have already demonstrated immune system activation through peripheral blood flow cytometry, with decreased T-regulatory cells and down-regulation of PD-1 expression in T cells after therapy. Given the limited sample size and the overall high pCR rate, we were unable to identify specific patterns associated with resistance to both immunotherapy and radiotherapy. Regarding the gut microbiome, we found that Akkermansia, Bifidobacterium, and unclassified Ruminococcaceae were more associated with patients who had a complete pathologic response, compared with those who did not respond to treatment. This result is in concordance with other works that also observed that patients with a good response to immunotherapy presented an enrichment of Akkermansia muciniphila and Bifidobacterium in their stools.39 It is worth highlighting that the sole patient displaying no evidence of pathologic regression (with 100% viable tumor cells) harbored an EGFR mutation, a low PD-L1 expression, and a reduction of bacterial genera associated with better responses to immunotherapy. This case underscores that certain molecular and microenvironmental features often linked to immunotherapy resistance may be mitigated by the addition of radiotherapy and surgery. Nevertheless, the resistance observed in this specific case could have been detected by the absence of any anatomical or metabolic regression in the post-treatment PET-CT, becoming the single case in which surgery certainly benefited the patient. Although radiologic findings tended to underestimate pathologic response overall, all other patients exhibited some radiologic changes indicative of treatment activity, suggesting that imaging could play a role in selecting patients who would indeed require surgical resection after neoadjuvant therapy.
Our study had several limitations. First, the single-center nature and the impact of the COVID-19 pandemic significantly restricted our sample size. Of the 25 recruited patients, 24 were from the public hospital affiliated with HIAE, representing an underserved population with substantial tobacco exposure (an average of 77 pack-years) and multiple comorbidities. Conducting this study during such challenging times was demanding, and we commend the dedication of all staff, patients, and caregivers involved. Second, despite robust preplanned translational procedures, the high pathologic response rate precluded any statistical conclusions and limited hypothesis generation to assess which patients benefit more from the nivolumab-SABR combination, who could be safely spared from surgery, or the evaluation of biomarkers of resistance. However, these results overall suggest that almost all patients achieved excellent cancer-specific outcomes and could be considered for noninvasive treatment in an all-comer fashion, rather than biomarker selected. This is particularly relevant because several patients died during follow-up from other comorbidities without cancer recurrence, emphasizing the need for less invasive/morbid treatments for patients at greater risk of death from other causes. Third, our follow-up period remains short, given that the primary end point was pathologic response, though all patients had at least 1 year of follow-up. Last, the lack of randomization to different cohorts, including testing various radiation doses and omitting surgery, limits the conclusions and underscores the need for future research in this area.
In conclusion, our study demonstrates a high pCR rate of 79% with the incorporation of three doses of nivolumab with SABR before surgery. Our findings offer important insights to guide the ongoing trial designs of combination strategies for early stage lung cancer—both in surgical candidates, where such approaches may serve to intensify treatment in high-risk stage I disease or potentially allow omission of surgery, and in patients who are inoperable.

CRediT Authorship Contribution Statement
Gustavo Schvartsman: Conceptualization, Supervision, Writing, Data curation, Investigation, Validation, Funding acquisition.
Yasmin Maria Amirato: Writing - review & editing, Validation.
Frederico Monfardini: Writing - review & editing, Formal analysis.
Gustavo Prado dos Santos: Writing - review & editing, Formal analysis.
Diogo Bugano Diniz Gomes: Writing - review & editing.
Ludmila de Oliveira Muniz Koch: Writing - review & editing.
Benoit Jacques Bibas: Writing - review & editing, Investigation.
Oswaldo Gomes Junior: Writing - review & editing, Investigation.
Paulo Vidal Campregher: Writing - review & editing, Methodology, Investigation.
Patrícia Severino: Writing - review & editing, Methodology, Investigation.
Luciana Cavalheiro Marti: Writing - review & editing, Methodology, Investigation.
Vitor Ribeiro Paes: Writing - review & editing, Methodology, Investigation.
Laura Leaden: Writing - review & editing.
Rodrigo Caruso Chate: Writing - review & editing, Methodology, Investigation.
Jose Milanez de Campos Ribas: Writing - review & editing.
Victor Aurélio Ramos Sousa: Writing - review & editing.
Patricia Taranto: Writing - review & editing.
Fernando Moura: Writing - review & editing.
Ricardo Mingarini Terra: Writing - review & editing.
Marcos Samano: Writing - review & editing, Investigation, Methodology.

Disclosure
Dr. Schvartsman declares receiving consulting fees from Bristol Myers Squibb, Merck, Sharp & Dome, Daiichi Sankyo, AstraZeneca, Sanofi, Roche, and Amgen. The remaining authors declare no conflict of interest.