RESUMEN

Objective: This study aims to examine whether radiation therapy doses are related to incidences of carotid artery stenosis and brain necrosis in a large-scale real-world database. Methods: We identified a cohort of HNC patients from the catastrophic illness patient dataset using ICD-9 or ICD-10 to compare the incidence and risks of carotid artery stenosis (CAS) and brain necrosis (RIBN) in patients who received a radiation therapy dose of ≥5400 cGy/30 fractions (group A) with those who received a radiation therapy dose of <5400 cGy/30 fractions (group B). The incidence and hazard ratios were quantified using Cox proportional hazards models. Results: A total of 19,964 patients were identified in group A and group B. Among them, 965 and 863 cases of CAS and 435 and 359 cases of RIBN were identified in group A and group B, respectively. There was no statistically significant association between the two groups for CAS risk, whereas there was a statistically significant association between the two groups for RIBN risk. The most common primary site of head and neck cancers was the nasopharynx (1144 of 19,964, 5.73%). Conclusions: Our study suggests that RT may increase the risk of carotid stenosis and brain necrosis in patients with NPC. To ensure patient safety during treatment, the optimal balance between tumor control and toxicity prevention in individual patients through minimization of the radiation dose to all relevant OARs must be properly understood.

RESUMEN

Mutation study for high-risk breast and ovarian cancer (HBOC) has been extensively studied in patients of different ethnicities. Here we compared the germline mutation rate and mutation spectrum of patients (n = 4341) with benign breast diseases or breast cancers, with and without other risk factors. Three cohorts of Chinese patients were recruited. The first cohort, high-risk cohort (HR, n = 3935) included high-risk breast cancer patients fulfilling high-risk HBOC criteria and who are recruited at our genetics clinic. The second cohort, unselected cancer cohort (CC, n = 307) was from general recruitment of patients with breast cancer at breast surgery clinics. The third cohort, benign breast lesion cohort (NC, n = 99) comprised 99 patients with benign breast diseases such as fibroadenoma, fibroadenomatoid hyperplasia, and intraductal papilloma. Thirty HBOC related genes were sequenced on the above-mentioned patient cohorts. The germline mutation rates of HR, CC, and NC cohort were 11.9%, 6.5%, and 8.1%, respectively. In the CC cohort, 29.3% (90/307) of patients fulfilled the National Comprehensive Cancer Network (NCCN) high-risk genetic test criteria 2022 v.2. The mutation rate for this group of patients was 11.1%, similar to that of the HR cohort, while the mutation rate for those not fulfilling testing criteria was 4.6%, like that of the NC cohort. High penetrance genes (BRCA1/2, CDH1, PALB2, PTEN, and TP53) mutations were only found in the HR (10.6%) and CC (3.3%) cohorts but were not found in the NC cohort. ATM, BRIP1, RAD51C, and RAD51D mutations were identified in all cohorts. RAD51C and RAD51D mutations showed conflicting penetrance. An unexpectedly high mutation rate of total 2% was found in the NC cohort but it was only 0.3% and 0.5% in the HR cohort and CC cohort, respectively. Our results show a clinical need to enhance genetic testing of unselected breast cancer patients to identify the high-risk patients.

RESUMEN

Multispectral image (MS) and panchromatic image (PAN) fusion, which is also named as multispectral pansharpening, aims to obtain MS with high spatial resolution and high spectral resolution. However, due to the usual neglect of noise and blur generated in the imaging and transmission phases of data during training, many deep learning (DL) pansharpening methods fail to perform on the dataset containing noise and blur. To tackle this problem, a variational optimization-guided two-stage network (VOGTNet) for multispectral pansharpening is proposed in this work, and the performance of variational optimization (VO)-based pansharpening methods relies on prior information and estimates of spatial-spectral degradation from the target image to other two original images. Concretely, we propose a dual-branch fusion network (DBFN) based on supervised learning and train it by using the datasets containing noise and blur to generate the prior fusion result as the prior information that can remove noise and blur in the initial stage. Subsequently, we exploit the estimated spectral response function (SRF) and point spread function (PSF) to simulate the process of spatial-spectral degradation, respectively, thereby making the prior fusion result and the adaptive recovery model (ARM) jointly perform unsupervised learning on the original dataset to restore more image details and results in the generation of the high-resolution MSs in the second stage. Experimental results indicate that the proposed VOGTNet improves pansharpening performance and shows strong robustness against noise and blur. Furthermore, the proposed VOGTNet can be extended to be a general pansharpening framework, which can improve the ability to resist noise and blur of other supervised learning-based pansharpening methods. The source code is available at https://github.com/HZC-1998/VOGTNet.

RESUMEN

BACKGROUND: We aimed to compare 7 newer immunotherapies and targeted therapies for platinum-resistant relapsed ovarian cancer. METHODS: We conducted a comprehensive search of PubMed, EMBASE, and Cochrane Library electronic databases for phase III trials involving platinum-resistant recurrent ovarian cancer (PRrOC) patients treated with immunotherapy or targeted therapy in combination with chemotherapy. The quality of the included trials was assessed using the GRADE method. The primary outcome of comparison was progression-free survival, and secondary outcomes included overall survival and safety. RESULTS: This analysis included 7 randomized phase III controlled trials, encompassing 2485 PRrOC patients. Combining bevacizumab plus chemotherapy and lurbinectedin demonstrated statistically significant differences in progression-free survival compared to all other regimens of interest. However, no statistically significant differences were observed in the overall survival. Nivolumab and mirvetuximab exhibited fewer serious adverse events than the other regimens of interest. CONCLUSIONS: Our findings indicate that bevacizumab combined with chemotherapy and lurbinectedin monotherapy has significant efficacy in patients with PRrOC. For patients with PRrOC who have exhausted treatment options, nivolumab and mirvetuximab may be considered as alternatives because of their better safety profiles.

Asunto(s)

RESUMEN

Background: Portal vein tumor thrombus (PVTT) is a common complication and an obstacle to treatment, with a high recurrence rate and poor prognosis. There is still no global consensus or standard guidelines on the management of hepatocellular carcinoma (HCC) with PVTT. Increasing evidence suggests that more aggressive treatment modalities, including transarterial chemoembolization, radiotherapy, targeted therapy, and various combination therapies, may improve the prognosis and prolong the survival of advanced hepatocellular carcinoma (aHCC) patients with PVTT. We aim to comprehensively review and compare the efficacy and safety of these advanced options for aHCC with PVTT. Methods: A comprehensive literature search was conducted on PubMed and EMBASE for phase II or III randomized controlled trials (RCTs) investigating multimodality treatments for aHCC with PVTT. Kaplan-Meier curves for overall survival (OS) and progression-free survival were constructed to retrieve individual patient-level data to strengthen the comparison of the benefits of all multimodality treatments of interest. Each study was pooled in a fixed-effects network meta-analysis (NMA). We also conducted subgroup analyses using risk ratios extracted from each study, including viral etiology, Barcelona Clinic Liver Cancer (BCLC) staging, alpha-fetoprotein (AFP) levels, macrovascular invasion or portal vein tumor thrombosis, and extrahepatic spread. Multimodality treatments were ranked using SUCRA scores. Results: We identified 15 randomized controlled trials with 16 multimodality regimens that met the inclusion criteria. Among them, 5,236 patients with OS results and 5,160 patients with PFS results were included in the analysis. The hepatic arterial infusion chemotherapy of fluorouracil, leucovorin, and oxaliplatin (HAIC-FO) showed OS and PFS benefits over all the other therapies. In terms of OS, HAIC-FO, nivolumab, and TACE+Len were superior to sorafenib, lenvatinib, and donatinib monotherapies, as well as HAIC-FO+Sor. In terms of PFS, TACE+Len showed better benefits than lenvatinib, donatinib, and tremelimumab+durvalumab. A low heterogeneity (I 2 < 50%) and consistency were observed. The SUCRA score for OS ranked HAIC-FO+sorafenib as the best treatment option among all multimodality treatments in hepatitis B, MVI, or PVTT with EHS and AFP 400 µg/L subgroups. Conclusion: HAIC-FO and HAIC-FO+sorafenib are statistically better options for unresectable hepatocellular carcinoma with PVTT among the multimodality treatments, and their effective and safe implementation may provide the best outcomes for HCC-PVTT patients.

RESUMEN

The purpose of this study was to assess the risks of hyperthyroidism and hypothyroidism related to gynecological cancers. Population-based retrospective cohort study. We conducted a cohort study using the Taiwan National Health Insurance Research Database to explore hyperthyroidism and hypothyroidism associated with site-specific gynecologic cancers in women from January 1, 2000 to December 31, 2018. The examined gynecologic cancers included endometrial (EC), uterine corpus cancer (UC), and ovarian cancer (OC). The incidence and hazard ratios were quantified using Cox proportional hazards models. The incidence of developing gynecological (Gyn) cancers in the hyperthyroid and hypothyroid women was 0.29 and 0.44 per 1000 person-years, which was 0.86 fold lower and 1.13 fold higher than that in the comparison cohort (p < 0.001). Compared with patients aged 20-40 years, patients in older age groups had a lower and higher risk of developing Gyn cancers (for hyperthyroid, 40-65 years: adjusted hazard ratio (aHR) = 0.82; > 65 years: aHR = 0.94; for hypothyroid, adjusted hazard ratio (aHR) = 1.26; > 65 years: aHR = 1.38). Compared with the non-hypothyroid women and non-hyperthyroid women beyond 6 years of follow-up, hypothyroid and hyperthyroid women showed decreased risk of Gyn cancers. Medication treatment for hyperthyroid and hypothyroid disease did not showed significant association in subgroup analyses (aHR = 0.99 and 0.80, respectively). Our results show that women with hyperthyroidism have a significantly reduced risk of gynecological cancers, whereas women with hypothyroidism have a slightly increased risk of gynecological cancers suggesting an association between thyroid function level and risk of gynecological cancers.

Asunto(s)

RESUMEN

An optimal energy scheduling strategy for integrated energy systems (IESs) can effectively improve the energy utilization efficiency and reduce carbon emissions. Due to the large-scale state space of IES caused by uncertain factors, it would be beneficial for the model training process to formulate a reasonable state-space representation. Thus, a condition knowledge representation and feedback learning framework based on contrastive reinforcement learning is designed in this study. Considering that different state conditions would bring inconsistent daily economic costs, a dynamic optimization model based on deterministic deep policy gradient is established, so that the condition samples can be partitioned according to the preoptimized daily costs. In order to represent the overall conditions on a daily basis and constrain the uncertain states in the IES environment, the state-space representation is constructed by a contrastive network considering the time dependence of variables. A Monte-Carlo policy gradient-based learning architecture is further proposed to optimize the condition partition and improve the policy learning performance. To verify the effectiveness of the proposed method, typical load operation scenarios of an IES are used in our simulations. The human experience strategies and state-of-the-art approaches are selected for comparisons. The results validate the advantages of the proposed approach in terms of cost effectiveness and ability to adapt in uncertain environments.

RESUMEN

Object detection is important in many applications, such as autonomous driving. While 2D images lack depth information and are sensitive to environmental conditions, 3D point clouds can provide accurate depth information and a more descriptive environment. However, sparsity is always a challenge in single-frame point cloud object detection. This paper introduces a two-stage proposal-based feature fusion method for object detection using multiple frames. The proposed method, called proposal features fusion (PFF), utilizes a cosine-similarity approach to associate proposals from multiple frames and employs an attention weighted fusion (AWF) module to merge features from these proposals. It allows for feature fusion specific to individual objects and offers lower computational complexity while achieving higher precision. The experimental results on the nuScenes dataset demonstrate the effectiveness of our approach, achieving an mAP of 46.7%, which is 1.3% higher than the state-of-the-art 3D object detection method.

RESUMEN

Deformable image registration is a process to determine the non-linear spatial correspondence among deformed image pairs. Generative registration network is a novel structure involving a generative registration network and a discriminative network that encourages the former to generate better results. We propose an Attention Residual UNet (AR-UNet) to estimate the complicated deformation field. The model is trained using perceptual cyclic constraints. As an unsupervised method, we require labelling for training and use virtual data augmentation to improve the robustness of the proposed model. We also introduce comprehensive metrics for image registration comparison. Experimental results show quantitative evidence that the proposed method can predict reliable deformation field at a reasonable speed and outperform conventional learning based and non-learning based deformable image registration methods.

RESUMEN

As the pixel resolution of imaging equipment has grown larger, the images' sizes and the number of pixels used to represent objects in images have increased accordingly, exposing an issue when dealing with larger images using the traditional deep learning models and methods, as they typically employ mechanisms such as increasing the models' depth, which, while suitable for applications that have to be spatially invariant, such as image classification, causes issues for applications that relies on the location of the different features within the images such as object localization and change detection. This paper proposes an adaptive convolutional kernels layer (AKL) as an architecture that adjusts dynamically to images' sizes in order to extract comparable spectral information from images of different sizes, improving the features' spatial resolution without sacrificing the local receptive field (LRF) for various image applications, specifically those that are sensitive to objects and features locations, using the definition of Fourier transform and the relation between spectral analysis and convolution kernels. The proposed method is then tested using a Monte Carlo simulation to evaluate its performance in spectral information coverage across images of various sizes, validating its ability to maintain coverage of a ratio of the spectral domain with a variation of around 20% of the desired coverage ratio. Finally, the AKL is validated for various image applications compared to other architectures such as Inception and VGG, demonstrating its capability to match Inception v4 in image classification applications, and outperforms it as images grow larger, up to a 30% increase in accuracy in object localization for the same number of parameters.

RESUMEN

Purpose: To determine mortality and outcomes of patients diagnosed with fracture-related infections (FRIs).Methods: FRI patients treated at a trauma centre between 2001 and 2020 were analysed. The primary outcome was 1-year mortality; mortality associations with FRI organism, depth of involvement, and temporality were investigated with multivariable survival analysis. Healthcare-associated and serological outcomes were reported as secondary outcomes. Results: 311 FRIs with mean age of 67.0 and median Charlson comorbidity index of 0 were analysed. Methicillin-sensitive Staphylococcus aureus (MSSA) (29.9%) was the most frequently implicated organism. The majority of FRIs were deep infections (62.7%). FRIs were diagnosed at a median of 40 (IQR 15-200) days post index surgery. The mean follow-up was 5.9 years. One-year mortality amounted to 17.7%. MSSA FRIs were associated with better survival (adj HR 0.34, 95%CI 0.15-0.76, p = 0.008). There was no difference in survivorship between deep or superficial FRI (adj HR 0.86, 95%CI 0.62-1.19, p = 0.353) or in relation to onset time (adj HR 1.0, 95%CI 0.99-1.00, p = 0.943). Implant removal or debridement alone was performed in 61.7% and 17% respectively. Antibiotics was prescribed for 53 (IQR 23-110) days, and patients were hospitalised for 39 (IQR 19-78) days. CRP and ESR normalised in 70.3% (median 46 days) and 53.8% (median 86 days) patients respectively. Conclusion: Fracture-related infections are associated with significant mortality and morbidity regardless of depth and temporality. Non-MSSA FRIs are associated with inferior survival.

Asunto(s)

RESUMEN

Structural variation (SV) is a major cause of genetic disorders. In this paper, we show that low-depth (specifically, 4×) whole-genome sequencing using a single Oxford Nanopore MinION flow cell suffices to support sensitive detection of SV, particularly pathogenic SV for supporting clinical diagnosis. When using 4× ONT WGS data, existing SV calling software often fails to detect pathogenic SV, especially in the form of long deletion, terminal deletion, duplication, and unbalanced translocation. Our new SV calling software SENSV can achieve high sensitivity for all types of SV and a breakpoint precision typically ± 100 bp; both features are important for clinical concerns. The improvement achieved by SENSV stems from several new algorithms. We evaluated SENSV and other software using both real and simulated data. The former was based on 24 patient samples, each diagnosed with a genetic disorder. SENSV found the pathogenic SV in 22 out of 24 cases (all heterozygous, size from hundreds of kbp to a few Mbp), reporting breakpoints within 100 bp of the true answers. On the other hand, no existing software can detect the pathogenic SV in more than 10 out of 24 cases, even when the breakpoint requirement is relaxed to ± 2000 bp.

Asunto(s)

RESUMEN

BACKGROUND: The application of long-read sequencing using the Oxford Nanopore Technologies (ONT) MinION sequencer is getting more diverse in the medical field. Having a high sequencing error of ONT and limited throughput from a single MinION flowcell, however, limits its applicability for accurate variant detection. Medical exome sequencing (MES) targets clinically significant exon regions, allowing rapid and comprehensive screening of pathogenic variants. By applying MES with MinION sequencing, the technology can achieve a more uniform capture of the target regions, shorter turnaround time, and lower sequencing cost per sample. METHOD: We introduced a cost-effective optimized workflow, ECNano, comprising a wet-lab protocol and bioinformatics analysis, for accurate variant detection at 4800 clinically important genes and regions using a single MinION flowcell. The ECNano wet-lab protocol was optimized to perform long-read target enrichment and ONT library preparation to stably generate high-quality MES data with adequate coverage. The subsequent variant-calling workflow, Clair-ensemble, adopted a fast RNN-based variant caller, Clair, and was optimized for target enrichment data. To evaluate its performance and practicality, ECNano was tested on both reference DNA samples and patient samples. RESULTS: ECNano achieved deep on-target depth of coverage (DoC) at average > 100× and > 98% uniformity using one MinION flowcell. For accurate ONT variant calling, the generated reads sufficiently covered 98.9% of pathogenic positions listed in ClinVar, with 98.96% having at least 30× DoC. ECNano obtained an average read length of 1000 bp. The long reads of ECNano also covered the adjacent splice sites well, with 98.5% of positions having ≥ 30× DoC. Clair-ensemble achieved > 99% recall and accuracy for SNV calling. The whole workflow from wet-lab protocol to variant detection was completed within three days. CONCLUSION: We presented ECNano, an out-of-the-box workflow comprising (1) a wet-lab protocol for ONT target enrichment sequencing and (2) a downstream variant detection workflow, Clair-ensemble. The workflow is cost-effective, with a short turnaround time for high accuracy variant calling in 4800 clinically significant genes and regions using a single MinION flowcell. The long-read exon captured data has potential for further development, promoting the application of long-read sequencing in personalized disease treatment and risk prediction.

Asunto(s)

RESUMEN

HKG is the first fully accessible variant database for Hong Kong Cantonese, constructed from 205 novel whole-exome sequencing data. There has long been a research gap in the understanding of the genetic architecture of southern Chinese subgroups, including Hong Kong Cantonese. HKG detected 196 325 high-quality variants with 5.93% being novel, and 25 472 variants were found to be unique in HKG compared to three Chinese populations sampled from 1000 Genomes (CHN). PCA illustrates the uniqueness of HKG in CHN, and the admixture study estimated the ancestral composition of HKG and CHN, with a gradient change from north to south, consistent with their geological distribution. ClinVar, CIViC and PharmGKB annotated 599 clinically significant variants and 360 putative loss-of-function variants, substantiating our understanding of population characteristics for future medical development. Among the novel variants, 96.57% were singleton and 6.85% were of high impact. With a good representation of Hong Kong Cantonese, we demonstrated better variant imputation using reference with the addition of HKG data, thus successfully filling the data gap in southern Chinese to facilitate the regional and global development of population genetics.

RESUMEN

This article proposes the problem of joint state estimation and correlation identification for data fusion with unknown and time-varying correlation under the Bayesian learning framework. The considered data correlation is represented by the randomly weighted sum of positive semi-definite matrices, where the random weights depict at least three kinds of unknown correlation across single-sensor measurement components, multisensor measurements, and local estimates. Based on the variational Bayesian mechanism, the joint posterior distribution of the state and weights is derived in a closed-form iterative manner, through minimizing the Kullback-Leibler divergence. The three-case simulation shows the superiority of the proposed method in the root-mean-square error of estimation and identification.

RESUMEN

This article introduces a novel consensus-based labeled multi-Bernoulli (LMB) filter to tackle multitarget tracking (MTT) in a distributed sensor network (DSN), whose sensor nodes have limited and different fields of view (FoVs). Although consensus-based algorithms are effective for distributed fusion and MTT, it may be problematic when distributed sensor nodes have different FoVs. To deal with this issue, the proposed method constructs an extended label space mapping to overcome the "label space mismatching" phenomenon; after that, the model of the undetected multitargets is established so that the tracks can be initialized outside the FoV of local sensors; finally and most important, weight selection and evolution mechanism are proposed such that the fusion weights are automatically tuned for each track at each time step and consensus step. The efficiency and robustness of the proposed algorithm are demonstrated in a distributed MTT scenario via numerical simulations.

Asunto(s)

RESUMEN

BACKGROUND: Disease-drug associations provide essential information for drug discovery and disease treatment. Many disease-drug associations remain unobserved or unknown, and trials to confirm these associations are time-consuming and expensive. To better understand and explore these valuable associations, it would be useful to develop computational methods for predicting unobserved disease-drug associations. With the advent of various datasets describing diseases and drugs, it has become more feasible to build a model describing the potential correlation between disease and drugs. RESULTS: In this work, we propose a new prediction method, called LMFDA, which works in several stages. First, it studies the drug chemical structure, disease MeSH descriptors, disease-related phenotypic terms, and drug-drug interactions. On this basis, similarity networks of different sources are constructed to enrich the representation of drugs and diseases. Based on the fused disease similarity network and drug similarity network, LMFDA calculated the association score of each pair of diseases and drugs in the database. This method achieves good performance on Fdataset and Cdataset, AUROCs were 91.6% and 92.1% respectively, higher than many of the existing computational models. CONCLUSIONS: The novelty of LMFDA lies in the introduction of multimodal fusion using low-rank tensors to fuse multiple similar networks and combine matrix complement technology to predict potential association. We have demonstrated that LMFDA can display excellent network integration ability for accurate disease-drug association inferring and achieve substantial improvement over the advanced approach. Overall, experimental results on two real-world networks dataset demonstrate that LMFDA able to delivers an excellent detecting performance. Results also suggest that perfecting similar networks with as much domain knowledge as possible is a promising direction for drug repositioning.

Asunto(s)

RESUMEN

Background and Objectives: To evaluate the effectiveness of hepatic arterial infusion chemotherapy (HAIC) followed by lipiodol infusion in advanced hepatocellular carcinoma (HCC) with portal vein tumor thrombus (PVTT). Materials and Methods: Thirty-two patients with advanced HCC and PVTT who received HAIC with regimens of cisplatin, mitomycin-C, and 5-fluorouracil followed by lipiodol infusion were enrolled. The primary efficacy endpoint was tumor response rate. The modified Response Evaluation Criteria in Solid Tumors (mRECIST) was used for assessment of treatment response. The secondary endpoints were overall survival (OS) and progression free survival (PFS). Prognostic factors for survival also were evaluated. Results: The median OS and PFS were 11.9 and 9.5 months, respectively. Seventeen patients (53.1%) achieved objective response, and 23 patients (71.9%) achieved disease control. The length of survival in the responder and disease control groups was longer than in the non-responder and progressive disease groups after two cycles of HAIC (responder vs. non-responder: 16.5 vs. 7.9 months, p = 0.001; disease control vs. progressive disease: 12.3 vs. 5.6 months, p < 0.001) and after completing HAIC (responder vs. non-responder: 15.7 vs. 6.9 months, p = 0.001; disease control vs. progressive disease: 13.6 vs. 6.9 months, p < 0.001). Better survival was associated with Child-Pugh A liver function (p = 0.013), with early response to two HAIC cycles (p = 0.009), and with response (p = 0.02) and disease control (p = 0.001) after completing HAIC treatment. Conclusion: HAIC followed by lipiodol infusion is a safe and feasible treatment for advanced HCC with PVTT. Patients with early response could continue HAIC treatment with expected prolonged survival.

Asunto(s)

RESUMEN

Symbiotic autonomous systems (SAS) are advanced intelligent and cognitive systems that exhibit autonomous collective intelligence enabled by coherent symbiosis of human-machine interactions in hybrid societies. Basic research in the emerging field of SAS has triggered advanced general-AI technologies that either function without human intervention or synergize humans and intelligent machines in coherent cognitive systems. This work presents a theoretical framework of SAS underpinned by the latest advances in intelligence, cognition, computer, and system sciences. SAS are characterized by the composition of autonomous and symbiotic systems that adopt bio-brain-social-inspired and heterogeneously synergized structures and autonomous behaviours. This paper explores the cognitive and mathematical foundations of SAS. The challenges to seamless human-machine interactions in a hybrid environment are addressed. SAS-based collective intelligence is explored in order to augment human capability by autonomous machine intelligence towards the next generation of general AI, cognitive computers, and trustworthy mission-critical intelligent systems. Emerging paradigms and engineering applications of SAS are elaborated via autonomous knowledge learning systems that symbiotically work between humans and cognitive robots. This article is part of the theme issue 'Towards symbiotic autonomous systems'.

RESUMEN

BACKGROUND: We compared the efficacy and safety of combinations of cyclin-dependent kinase 4/6 (CDK4/6) inhibitors and PI3K/AKT/mTOR inhibitors as second-line treatment in postmenopausal women with HR+, HER2- metastatic breast cancer. METHODS: We searched the Medline, Embase, and Cochrane Library electronic databases for phase II/III randomized trials evaluating CDK4/6 and PI3K/AKT/mTOR inhibitors plus fulvestrant. We compared the results with a network meta-analysis. Study quality was assessed following the GRADE approach. Outcomes of interest were progression-free survival, overall response rate, overall survival and G3-4 adverse drug events (ADEs). RESULTS: Eight RCTs were identified in the network meta-analysis. PFS was significantly improved by treatment with abemaciclib plus fulvestrant and ribociclib plus fulvestrant compared to pictilisib plus fulvestrant. The ORR following treatment with abemaciclib plus fulvestrant, ribociclib plus fulvestrant, palbociclib plus fulvestrant, buparlisib plus fulvestrant, and alpelisib plus fulvestrant significantly differed from that observed following treatment with placebo plus fulvestrant. In terms of OS, compared with placebo plus fulvestrant, abemaciclib plus fulvestrant, ribociclib plus fulvestrant, and buparlisib plus fulvestrant had a significant difference. The risks of ADEs were similar among three CDK4/6 inhibitors. CONCLUSION: As second-line treatment, three CDK4/6 inhibitors showed superior clinical efficacy compared to other PI3K/AKT/mTOR inhibitors with comparable safety profiles.

Asunto(s)