Family-wise Error Research Articles

Adaptive designs for IVPT data with mixed scaled average bioequivalence.

In vitro permeation tests (IVPT) offer accurate and cost-effective development pathways for locally acting drugs, such as topical dermatological products. For assessment of bioequivalence, the FDA draft guidance on generic acyclovir 5% cream introduces a new experimental design, namely the single-dose, multiple-replicate per treatment group design, as IVPT pivotal study design. We examine the statistical properties of its hypothesis testing method-namely the mixed scaled average bioequivalence (MSABE). Meanwhile, some adaptive design features in clinical trials can help researchers make a decision earlier with fewer subjects or boost power, saving resources, while controlling the impact on family-wise error rate. Therefore, we incorporate MSABE in an adaptive design combining the group sequential design and sample size re-estimation. Simulation studies are conducted to study the passing rates of the proposed methods-both within and outside the average bioequivalence limits. We further consider modifications to the adaptive designs applied for IVPT BE trials, such as Bonferroni's adjustment and conditional power function. Finally, a case study with real data demonstrates the advantages of such adaptive methods.

Optimal test procedures for multiple hypotheses controlling the familywise expected loss.

We consider the problem of testing multiple null hypotheses, where a decision to reject or retain must be made for each one and embedding incorrect decisions into a real-life context may inflict different losses. We argue that traditional methods controlling the Type I error rate may be too restrictive in this situation and that the standard familywise error rate may not be appropriate. Using a decision-theoretic approach, we define suitable loss functions for a given decision rule, where incorrect decisions can be treated unequally by assigning different loss values. Taking expectation with respect to the sampling distribution of the data allows us to control the familywise expected loss instead of the conventional familywise error rate. Different loss functions can be adopted, and we search for decision rules that satisfy certain optimality criteria within a broad class of decision rules for which the expected loss is bounded by a fixed threshold under any parameter configuration. We illustrate the methods with the problem of establishing efficacy of a new medicinal treatment in non-overlapping subgroups ofpatients.

Multiple sclerosis lesions that impair memory map to a connected memory circuit.

Nearly 1 million Americans are living with multiple sclerosis (MS) and 30-50% will experience memory dysfunction. It remains unclear whether this memory dysfunction is due to overall white matter lesion burden or damage to specific neuroanatomical structures. Here we test if MS memory dysfunction is associated with white matter lesions to a specific brain circuit. We performed a cross-sectional analysis of standard structural images and verbal memory scores as assessed by immediate recall trials from 431 patients with MS (mean age 49.2 years, 71.9% female) enrolled at a large, academic referral center. White matter lesion locations from each patient were mapped using a validated algorithm. First, we tested for associations between memory dysfunction and total MS lesion volume. Second, we tested for associations between memory dysfunction and lesion intersection with an a priori memory circuit derived from stroke lesions. Third, we performed mediation analyses to determine which variable was most associated with memory dysfunction. Finally, we performed a data-driven analysis to derive de-novo brain circuits for MS memory dysfunction using both functional (n = 1000) and structural (n = 178) connectomes. Both total lesion volume (r = 0.31, p < 0.001) and lesion damage to our a priori memory circuit (r = 0.34, p < 0.001) were associated with memory dysfunction. However, lesion damage to the memory circuit fully mediated the association of lesion volume with memory performance. Our data-driven analysis identified multiple connections associated with memory dysfunction, including peaks in the hippocampus (T = 6.05, family-wise error p = 0.000008), parahippocampus, fornix and cingulate. Finally, the overall topography of our data-driven MS memory circuit matched our a priori stroke-derived memory circuit. Lesion locations associated with memory dysfunction in MS map onto a specific brain circuit centered on the hippocampus. Lesion damage to this circuit fully mediated associations between lesion volume and memory. A circuit-based approach to mapping MS symptoms based on lesions visible on standard structural imaging may prove useful for localization and prognosis of higher order deficits in MS.

Improved family-wise error rate control in multiple equivalence testing.

Equivalence testing is an important component of safety assessments, used for example by the European Food Safety Authority, to allow new food or feed products on the market. The aim of such tests is to demonstrate equivalence of characteristics of test and reference crops. Equivalence tests are typically univariate and applied to each measured analyte (characteristic) separately without multiplicity correction. This increases the probability of making false claims of equivalence (type I errors) when evaluating multiple analytes simultaneously. To solve this problem, familywise error rate (FWER) control using Hochberg's method has been proposed. This paper demonstrates that, in the context of equivalence testing, other FWER-controlling methods are more powerful than Hochberg's. Particularly, it is shown that Hommel's method is guaranteed to perform at least as well as Hochberg's and that an "adaptive" version of Bonferroni's method, which uses an estimator of the proportion of non-equivalent characteristics, often substantially outperforms Hommel's method. Adaptive Bonferroni takes better advantage of the particular context of food safety where a large proportion of true equivalences is expected, a situation where other methods are particularly conservative. The different methods are illustrated by their application to two compositional datasets and further assessed and compared using simulated data.

A False Discovery Rate approach to optimal volatility forecasting model selection

Estimating financial market volatility is integral to the study of investment decisions and behaviour. Previous literature has, therefore, attempted to identify an optimal volatility forecasting model. However, optimal volatility forecasting is dynamic. It depends on the asset being studied and financial market conditions. We propose a novel empirical methodology to account for this dynamism. Using our Multiple Hypothesis Testing with the False Discovery Rate (FDR) method, we identify buckets of superior-performing models relative to the literature’s benchmark models. We present evidence that our proposed FDR bucket with GJR-GARCH has the lowest forecast error in predicting one-step-ahead realized volatility. We also compare our FDR method with two Family-Wise Error Rate model selection frameworks, and the evidence supports our proposed FDR methodology.

False discovery rate envelopes

False discovery rate (FDR) is a common way to control the number of false discoveries in multiple testing. There are a number of approaches available for controlling FDR. However, for functional test statistics, which are discretized into m highly correlated hypotheses, the methods must account for changes in distribution across the functional domain and correlation structure. Further, it is of great practical importance to visualize the test statistic together with its rejection or acceptance region. Therefore, the aim of this paper is to find, based on resampling principles, a graphical envelope that controls FDR and detects the outcomes of all individual hypotheses by a simple rule: the hypothesis is rejected if and only if the empirical test statistic is outside of the envelope. Such an envelope offers a straightforward interpretation of the test results, similarly as the recently developed global envelope testing which controls the family-wise error rate. Two different adaptive single threshold procedures are developed to fulfill this aim. Their performance is studied in an extensive simulation study. The new methods are illustrated by three real data examples.

An Improved Dunnett’s Procedure for Comparing Multiple Treatments with a Control in the Presence of Missing Observations

Dunnett’s procedure has been frequently used for multiple comparisons of group means of several treatments with a control, in drug development and other areas. However, in practice, researchers usually face missing observations when performing Dunnett’s procedure. This paper presents an improved Dunnett’s procedure that can construct unique ensemble confidence intervals for comparing group means of several treatments with a control, in the presence of missing observations, using a derived multivariate t distribution under the framework of Rubin’s rule. This procedure fills the current research gap that Rubin’s repeated-imputation inferences cannot adjust for multiplicity and, thereby, cannot give a unified confidence interval to control the family-wise error rate (FWER) when dealing with this problem. Simulation results show that the constructed pooled confidence intervals archive nominal joint coverage and the interval estimations preserve comparable precision to Rubin’s repeated-imputation inference as the missing rate increases. The proposed procedure with propensity-score imputation method is shown to produce more accurate interval estimations and control the FWER well.

Multiple-hypothesis testing rules for high-dimensional model selection and sparse-parameter estimation

We consider the problem of model selection for high-dimensional sparse linear regression models. We pose the model selection problem as a multiple-hypothesis testing problem and employ the methods of false discovery rate (FDR) and familywise error rate (FER) to solve it. We also present the reformulation of the FDR/FER-based approaches as criterion-based model selection rules and establish their relation to the extended Bayesian Information Criterion (EBIC), which is a state-of-the-art high-dimensional model selection rule. We use numerical simulations to show that the proposed FDR/FER method is well suited for high-dimensional model selection and performs better than EBIC.

Magnetic susceptibility changes in the brainstem reflect REM sleep without atonia severity in isolated REM sleep behavior disorder

REM sleep without atonia (RWA) is the hallmark of isolated REM sleep behavior disorder (iRBD) and is caused by neurodegeneration of brainstem structures. Previously, quantitative susceptibility mapping (QSM) was shown to detect microstructural tissue changes in neurodegenerative diseases. The goal of the study was to compare brainstem magnetic susceptibility (MS) in iRBD and controls using the voxel-based QSM approach and to examine the association between brainstem MS and severity of RWA in iRBD. Sixty iRBD patients and 41 healthy controls were included in the study. Phasic, tonic, mixed RWA and SINBAR score was quantified. QSM maps were reconstructed with QSMbox software from a multi-gradient-echo sequence acquired at 3T MRI system and normalized using a custom T1 template. Voxel-based analysis with age and gender as covariates was performed using a two-sample t-test model for between-group comparison and using a linear regression model for association with the RWA parameters. Statistical maps were generated using threshold free cluster enhancement with p-value p < 0.05, corrected for family wise error. Compared to controls, the iRBD group had higher MS in bilateral substantia nigra (SN), red nucleus and the ventral tegmental area. MS positively correlated with iRBD duration in the right pedunculotegmental nucleus and white matter of caudal mesencephalic and pontine tegmentum and with phasic RWA in bilateral SN. QSM was able to detect MS abnormalities in several brainstem structures in iRBD. Association of MS levels in the brainstem with the intensity of RWA suggests that increased iron content in SN is related to RWA severity.

Properties of a Confirmatory Two-Stage Adaptive Procedure for Assessing Average Bioequivalence

We investigate a confirmatory two-stage adaptive procedure for assessing average bioequivalence and provide some insights to its theoretical properties. Effectively, we perform Two One-Sided Tests (TOST) to reach an overall decision about each of the two traditional null-hypotheses involved in declaring average bioequivalence. The tests are performed as combination tests separately for each hypothesis based on the corresponding pair of stagewise p-values. Features of the procedure include a built in futility, sample size reassessment, and the ability to simultaneously assess average bioequivalence with respect to multiple endpoints while controlling the familywise error rate. To facilitate inference at the end of a trial we consider confidence limits that match the decision reached on each one sided hypothesis and provide theory ensuring their appropriateness. The performance is assessed by simulation in the context of planning a study to compare two different administrations of an antibody treatment of atopicdermatitis.

A neutral comparison of algorithms to minimize L0 penalties for high-dimensional variable selection.

Variable selection methods based on L0 penalties have excellent theoretical properties to select sparse models in a high-dimensional setting. There exist modifications of the Bayesian Information Criterion (BIC) which either control the familywise error rate (mBIC) or the false discovery rate (mBIC2) in terms of which regressors are selected to enter a model. However, the minimization of L0 penalties comprises a mixed-integer problem which is known to be NP-hard and therefore becomes computationally challenging with increasing numbers of regressor variables. This is one reason why alternatives like the LASSO have become so popular, which involve convex optimization problems that are easier to solve. The last few years have seen some real progress in developing new algorithms to minimize L0 penalties. The aim of this article is to compare the performance of these algorithms in terms of minimizing L0 -based selection criteria. Simulation studies covering a wide range of scenarios that are inspired by genetic association studies are used to compare the values of selection criteria obtained with different algorithms. In addition, some statistical characteristics of the selected models and the runtime of algorithms are compared. Finally, the performance of the algorithms is illustrated in a real data example concerned with expression quantitative trait loci (eQTL) mapping.

Cluster extent inference revisited: quantification and localisation of brain activity

Abstract Cluster inference based on spatial extent thresholding is a popular analysis method multiple testing in spatial data, and is frequently used for finding activated brain areas in neuroimaging. However, the method has several well-known issues. While powerful for finding regions with some activation, the method as currently defined does not allow any further quantification or localisation of signal. In this paper, we repair this gap. We show that cluster-extent inference can be used (1) to infer the presence of signal in any region of interest and (2) to quantify the percentage of activation in such regions. These additional inferences come for free, i.e. they do not require any further adjustment of the alpha-level of tests, while retaining full family-wise error control. We achieve this extension of the possibilities of cluster inference by embedding the method into a closed testing procedure, and solving the graph-theoretic k-separator problem that results from this embedding. We demonstrate the usefulness of the improved method in a large-scale application to neuroimaging data from the Neurovault database.

Abnormal cerebral blood flow in patients with Leber's hereditary optic neuropathy.

The study aimed to unravel abnormal cerebral blood flow (CBF) in patients with Leber's hereditary optic neuropathy (LHON) using arterial spin labeling (ASL) and to investigate the associations among disrupted CBF, disease duration, and neuro-ophthalmological impairment. ASL perfusion imaging data was collected from 20 patients with acute LHON, 29 patients with chronic LHON, and 37 healthy controls. We used a one-way analysis of covariance to test the intergroup differences in CBF. Linear and nonlinear curve fit models were applied to explore the associations among CBF, disease duration, and neuro-ophthalmological metrics. Brain regions differed in LHON patients, including the left sensorimotor and bilateral visual areas (p < 0.05, cluster-wise family-wise error correction). Acute and chronic LHON patients demonstrated lower CBF in bilateral calcarine than the healthy controls. Chronic LHON had lower CBF in the left middle frontal gyrus and sensorimotor cortex, and temporal-partial junction than the healthy controls and acute LHON. A significant logarithmic negative correlation was shown between CBF of left middle frontal gyrus and disease duration. A significant linear positive correlation was found between retinal nerve fiber layer thickness and CBF in left middle frontal gyrus, and negative correlations between loss of variance and CBF in left middle frontal gyrus and sensorimotor cortex (p < 0.05, Bonferroni correction). LHON patients exhibited reduced CBF in the visual pathway, sensorimotor and higher-tier cognitive areas. Disease duration and neuro-ophthalmological impairments can influence the metabolism of non-visual areas.

Permutation-based multiple testing corrections for -values and confidence intervals for cluster randomized trials.

In this article, we derive and compare methods to derive P-values and sets of confidence intervals with strong control of the family-wise error rates and coverage for estimates of treatment effects in cluster randomized trials with multiple outcomes. There are few methods for P-value corrections and deriving confidence intervals, limiting their application in this setting. We discuss the methods of Bonferroni, Holm, and Romano and Wolf and adapt them to cluster randomized trial inference using permutation-based methods with different test statistics. We develop a novel search procedure for confidence set limits using permutation tests to produce a set of confidence intervals under each method of correction. We conduct a simulation-based study to compare family-wise error rates, coverage of confidence sets, and the efficiency of each procedure in comparison to no correction using both model-based standard errors and permutation tests. We show that the Romano-Wolf type procedure has nominal error rates and coverage under non-independent correlation structures and is more efficient than the other methods in a simulation-based study. We also compare results from the analysis of a real-world trial.

Shared genetic risk loci between Alzheimer’s disease and related dementias, Parkinson’s disease, and amyotrophic lateral sclerosis

BackgroundGenome-wide association studies (GWAS) have indicated moderate genetic overlap between Alzheimer’s disease (AD) and related dementias (ADRD), Parkinson’s disease (PD) and amyotrophic lateral sclerosis (ALS), neurodegenerative disorders traditionally considered etiologically distinct. However, the specific genetic variants and loci underlying this overlap remain almost entirely unknown.MethodsWe leveraged state-of-the-art GWAS for ADRD, PD, and ALS. For each pair of disorders, we examined each of the GWAS hits for one disorder and tested whether they were also significant for the other disorder, applying Bonferroni correction for the number of variants tested. This approach rigorously controls the family-wise error rate for both disorders, analogously to genome-wide significance.ResultsEleven loci with GWAS hits for one disorder were also associated with one or both of the other disorders: one with all three disorders (the MAPT/KANSL1 locus), five with ADRD and PD (near LCORL, CLU, SETD1A/KAT8, WWOX, and GRN), three with ADRD and ALS (near GPX3, HS3ST5/HDAC2/MARCKS, and TSPOAP1), and two with PD and ALS (near GAK/TMEM175 and NEK1). Two of these loci (LCORL and NEK1) were associated with an increased risk of one disorder but decreased risk of another. Colocalization analysis supported a shared causal variant between ADRD and PD at the CLU, WWOX, and LCORL loci, between ADRD and ALS at the TSPOAP1 locus, and between PD and ALS at the NEK1 and GAK/TMEM175 loci. To address the concern that ADRD is an imperfect proxy for AD and that the ADRD and PD GWAS have overlapping participants (nearly all of which are from the UK Biobank), we confirmed that all our ADRD associations had nearly identical odds ratios in an AD GWAS that excluded the UK Biobank, and all but one remained nominally significant (p < 0.05) for AD.ConclusionsIn one of the most comprehensive investigations to date of pleiotropy between neurodegenerative disorders, we identify eleven genetic risk loci shared among ADRD, PD, and ALS. These loci support lysosomal/autophagic dysfunction (GAK/TMEM175, GRN, KANSL1), neuroinflammation/immunity (TSPOAP1), oxidative stress (GPX3, KANSL1), and the DNA damage response (NEK1) as transdiagnostic processes underlying multiple neurodegenerative disorders.

Brain FDG-PET findings in chimeric antigen receptor T-cell therapy neurotoxicity for diffuse large B-cell lymphoma.

Chimeric antigen receptor (CAR) T-cell therapy is potentially associated with treatment-related toxicities mainly consisting of cytokine release syndrome (CRS) and immune-effector cell-associated neurotoxicity syndrome (ICANS). We evaluated brain metabolic correlates of CRS with and without ICANS in diffuse large B-cell lymphoma patients treated with CAR-T. Twenty-one refractory DLCBLs underwent whole-body and brain [18 F]-fluorodeoxyglucose (FDG) PET before and 30days after treatment with CAR-T. Five patients did not develop inflammatory-related side effects, 11 patients developed CRS, while in 5 patients CRS evolved in ICANS. Baseline and post-CAR-T brain FDG-PET were compared with a local controls dataset to identify hypometabolic patterns both at single-patient and group levels (p<.05 after correction for family-wise error [FWE). Metabolic tumor volume (MTV) and total lesion glycolysis (TLG) were computed on baseline FDG-PET and compared between patients' subgroups (t-test). ICANS showed an extended and bilateral hypometabolic pattern mainly involving the orbitofrontal cortex, frontal dorsolateral cortex, and anterior cingulate (p<.003 FWE-corrected). CRS without ICANS showed significant hypometabolism in less extended clusters mainly involving bilateral medial and lateral temporal lobes, posterior parietal lobes, anterior cingulate, and cerebellum (p<.002 FWE-corrected). When compared, ICANS showed a more prominent hypometabolism in the orbitofrontal and frontal dorsolateral cortex in both hemispheres than CRS (p<.002 FWE-corrected). Mean baseline MTV and TLG were significantly higher in ICANS than CRS (p<.02). Patients with ICANS are characterized by a frontolateral hypometabolic signature coherently with the hypothesis of ICANS as a predominant frontal syndrome and with the more prominent susceptibility of frontal lobes to cytokine-induced inflammation.

Novel insights into systemic sclerosis using a sensitive computational method to analyze whole-genome bisulfite sequencing data

BackgroundAbnormal DNA methylation is thought to contribute to the onset and progression of systemic sclerosis. Currently, the most comprehensive assay for profiling DNA methylation is whole-genome bisulfite sequencing (WGBS), but its precision depends on read depth and it may be subject to sequencing errors. SOMNiBUS, a method for regional analysis, attempts to overcome some of these limitations. Using SOMNiBUS, we re-analyzed WGBS data previously analyzed using bumphunter, an approach that initially fits single CpG associations, to contrast DNA methylation estimates by both methods.MethodsPurified CD4+ T lymphocytes of 9 SSc and 4 control females were sequenced using WGBS. We separated the resulting sequencing data into regions with dense CpG data, and differentially methylated regions (DMRs) were inferred with the SOMNiBUS region-level test, adjusted for age. Pathway enrichment analysis was performed with ingenuity pathway analysis (IPA). We compared the results obtained by SOMNiBUS and bumphunter.ResultsOf 8268 CpG regions of ≥ 60 CpGs eligible for analysis with SOMNiBUS, we identified 131 DMRs and 125 differentially methylated genes (DMGs; p-values less than Bonferroni-corrected threshold of 6.05–06 controlling family-wise error rate at 0.05; 1.6% of the regions). In comparison, bumphunter identified 821,929 CpG regions, 599 DMRs (of which none had ≥ 60 CpGs) and 340 DMGs (q-value of 0.05; 0.04% of all regions). The top ranked gene identified by SOMNiBUS was FLT4, a lymphangiogenic orchestrator, and the top ranked gene on chromosome X was CHST7, known to catalyze the sulfation of glycosaminoglycans in the extracellular matrix. The top networks identified by IPA included connective tissue disorders.ConclusionsSOMNiBUS is a complementary method of analyzing WGBS data that enhances biological insights into SSc and provides novel avenues of investigation into its pathogenesis.

Final analysis of multi-histology basket trial expansion of ado-trastuzumab emtansine in patients with HER2 amplified cancers.

3025 Background: Human epidermal growth factor receptor 2 ( HER2, ERBB2) amplification occurs in 5% of non-breast non-gastric solid tumors. Ado-trastuzumab emtansine (T-DM1) showed preliminary efficacy in patients with HER2 amplified lung, endometrial, salivary gland, biliary tract and ovarian cancers, but the extent of its differential effects across histologies is unknown. Methods: Patients with HER2 amplified solid tumors were enrolled and received treatment 3.6mg/kg IV every 3 weeks. The primary endpoint was overall response rate (ORR) using RECIST v1.1 or PERCIST. A basket trial expansion used a Simon two stage optimal design applied to each of 5 histology cohorts with type I error rate under 2%, power of 90%, H0 10%, H1 40%, with family-wise error rate at 10%. After first stage, lung, salivary gland and endometrial cohorts were expanded to 23 patients. The null hypothesis was rejected for each cohort separately, if at least 6 responses were observed in each cohort. Secondary endpoints included duration of response (DOR), progression-free survival (PFS) and toxicity. HER2 amplification was identified by fluorescence in situ hybridization (FISH), or next generation sequencing (NGS). Correlative studies were performed using tissue immunohistochemistry (IHC). Plasma cell-free DNA (cfDNA) was collected throughout study treatment. Results: 88 patients with 8 unique cancer types were treated across 5 cohorts of HER2 amplified lung, salivary gland, colorectal, endometrial and other cancers. The median age was 66 (26-90). Median line of prior therapy was 2 (1-7). ORR was 33% (29/87 including 11 CRs, 95% CI 24-44%), including 47% (9/19) for lung cancers, 87% (13/15, 8 CRs) for salivary gland cancers, 22% (5/23, 3 CRs) for endometrial cancers, 12% (1/8) for biliary cancers, 14% (1/7) for ovarian cancers. Median DOR was 9.7 months (95% CI 4.8, 20.2), median PFS was 2.76 months (95% CI 2.53, 5.39). There were 8 (9%) G3 treatment related toxicities. HER2 fold change corrected for purity and ploidy by FACETS algorithm correlated with response (p=0.04) and PFS&gt;=3 months (p=0.0037). HER2 amplification by NGS correlated with HER2/CEP17≥2 by FISH (67/71 tested) and IHC3+ (54/68 tested). We observed persistent HER2 amplification in plasma cfDNA during acquired resistance and progression on T-DM1 in patients with salivary gland cancers. Conclusions: Ado-trastuzumab emtansine showed promising efficacy in patients with HER2 amplified lung and salivary gland cancers as identified by NGS, meeting the primary endpoint. However, its efficacy did not meet prespecified response rate in patients with HER2 amplified endometrial, colorectal and other cancers. Histologic lineage differences in HER2 amplified cancers affect response and translational research is critical for further drug development. Clinical trial information: NCT02675829 .

On Familywise Error Rate Cutoffs under Pairwise Exchangeability

In a pairwise exchangeable dependence setting for test statistics, the cutoffs of Sarkar et al. (2016) may be viewed as a first iteration improvement of Holm (1979)’s classical cutoffs under a convexity condition on the copula. The cutoffs of Seneta and Chen (1997) which improve Holm’s in the present exchangeability setting, are shown, after an analogous first iteration step, to lead to a refinement of Sarkar et al. (2016). Further, we show that the convexity condition can be circumvented in practice, computationally. Improvement by iteration limit of cutoffs is considered for both procedures. Comparisons between the effects of the several cutoff sets are made by way of plots of the familywise error rate against correlation rho in the classic setting of the multivariate Normal; and the distributional setting of the multivariate Generalized Hyperbolic for the important Variance Gamma type subfamily, for which a convexity condition cannot be analytically verified.

Skin adverse events of anti-cancer treatments: An examination of drug-AE associations.

e18885 Background: Although anti-cancer treatments, including chemotherapy (CTs), targeted therapies (TTs), radiation therapy, and immunotherapy (ITs), are effective to treat cancer, they can be associated with significant skin-related toxicities (AEs). These AEs can cause discomfort and even lead to the discontinuation of therapies. However, a comprehensive estimation of the associations between cancer drug use and skin AEs is currently lacking. This study aimed to investigate these associations using a large dataset. Methods: This study utilized the US FDA Adverse Reporting System (FAERS) dataset, with a focus on HCPs reports from January 2013 to September 2022. The dataset consists of 3399830 reports involving 3084 all-field drugs and 16347 AEs. To minimize false positives, we employed a nearest-neighbor matching model to identify 10 control reports for each case report based on cosine distance on demographic and severity factors. To manage multiple comparisons and control the family-wise error rate (FWER) to 5%, we applied the Bonferroni correction. Results: A total of 146 marketed anti-cancer drugs were identified in the database with at least 5 reports of skin AEs. Out of the 2757 drug-AE pairs, 708 displayed a significant reporting odds ratio (ROR) greater than one, consisting of 102 drugs and 135 skin AEs. The minimum ROR was 1.25, and 50% of the associations displayed a ROR above 11. Rash was significantly associated with 44 drugs and dry skin with 25 drugs. Methotrexate was significantly associated with 35 different AEs and anti-BRAF vemurafenib with 26 AEs. TCs were present in 57% of the pairs, CTS in 38%, and immune checkpoint blocking agents in 5%. Multikinase-I were present in 15% of the pairs, followed by antimetabolites (14%). Considering the relative weight of skin AEs on the safety profile of the drugs, skin AE were present on average in 11% of the drug reports, with a maximum of 51% for mechlorethamine. 12% of reports concerning CTs contained a skin AE, 11% for TTs, and 8% for ITs. For CTs, 13% of reports about antimetabolites contained a skin AE, followed by taxanes (12%). Vinca-alkaloids and topoisomerase-I were less impacted with 6% of skin AE reports. For TCs, 23% of reports about EGFR inhibitors contained a skin AE, followed by BRAF inhibitors (21%). PARP inhibitors and BTK inhibitors were less impacted with 8% of skin AE reports. Conclusions: This study used a large dataset to examine the associations between cancer drugs and skin AEs. 146 anti-cancer drugs were found to have skin AEs, with rash and dry skin being the most reported AEs in the associations. TTs were most associated with skin AEs, followed by CTs. Methotrexate and vemurafenib had the most important number of associations. These data don’t allow evaluation of the incidence of skin AE effect with anti-cancer drugs as they are probably under-reported, but the findings emphasize the importance of monitoring skin AEs in patients exposed to anti-cancer treatments.