Citations

Clear

5212 citation(s) found for "Wang, Li-Ju".

Anatomic Drivers of J-Sign Presence and Severity: If There Is a Jump, Look for a Bump
Dandu, Navya , Phillips, Andrew , Haneberg, Erik , Elias, Tristan J. , Wang, Zachary , Trasolini, Nicholas A. , Yanke, Adam B. , Yanke, Adam B , Haneberg, Erik C , Hevesi, Mario

Mayo Clinic

The American Journal of Sports Medicine | March 12, 2025 | Volume: 53 | Issue: 5 | Pages: 1119-1126

10.1177/03635465251322788 | PMID 40071349 Get It

Background: Medial patellofemoral ligament reconstruction is frequently indicated for recurrent lateral patellar instability. The preoperative presence and severity of a J-sign have been associated with poorer postoperative outcomes. Purpose: To determine the underlying anatomic factors that contribute to the presence, severity, and jumping quality of the J-sign. Study Design: Cross-sectional study; Level of evidence, 3. Methods: All patients undergoing evaluation for patellar instability at a single institution between 2013 and 2023 and healthy controls without patellar instability were included. Patients with a history of knee osteotomies were excluded. The presence of a jumping J-sign and its relationship to patellofemoral measures including the Caton-Deschamps Index (CDI), trochlear dysplasia (Dejour grade), tibial tubercle–trochlear groove (TT-TG) distance, tibial tubercle lateralization, trochlear bump height, mechanical alignment, femoral anteversion, tibial torsion, trochlear medialization, patellar width, axial patellar/trochlear overlap, patellar height, trochlear height, and knee rotation angle (KRA) were measured using standardized 1.5-T magnetic resonance imaging (MRI). Univariate pairwise and multivariable analyses were performed to determine the factors associated with J-sign presence, severity, and quality. Results: Of the 130 knees with patellar instability, 89 (68.5%) demonstrated a J-sign on physical examination. In total, 44 (33.8%) patients demonstrated a 1-quadrant J-sign, 32 (24.6%) demonstrated a 2-quadrant smooth J-sign, and 13 (10.0%) demonstrated a jumping J-sign. A total of 22 control, noninstability cases were included. On multivariable analysis, increasing TT-TG distance (OR, 1.1 increase per millimeter; P = .04), external KRA (OR, 1.1 increase per degree; P = .02), and increasing CDI (OR, 1.3 increase per 0.1 increase in CDI; P = .02) were associated with J-sign presence. Increasing bump height (OR, 1.72 increase per millimeter; P = .007) and decreasing patellar width (OR, 0.89 decrease per millimeter; P = .076) were associated with a larger J-sign, when present. Increasing bump height (OR, 1.80 increase per millimeter; P = .018), increasing patellar width (OR, 1.33 increase per millimeter; P = .047), and decreasing CDI (OR, 0.009 decrease per 0.01 increase in ratio; P = .008) were associated with a jumping J-sign in comparison with a smooth 2-quadrant J-sign. A KRA of 10° (AUC, 0.70) and a cartilaginous bump height of 6.6 mm (AUC, 0.73) were thresholds associated with jumping J-sign presence. Conclusion: The presence of a J-sign is associated with MRI findings of relatively greater external tibiofemoral rotation, increased TT-TG distance, and increased patellar height, while J-sign severity and jumping quality are associated with the presence of additional underlying trochlear factors such as increased bump height. The anatomic drivers identified in this study should be further evaluated as possible factors associated with suboptimal outcomes after surgical management.
Copy Number Variation and Haplotype Analysis of 17q21.31 Reveals Increased Risk Associated with Progressive Supranuclear Palsy and Gene Expression Changes in Neuronal Cells
Wang, Hui , Chang, Timothy S. , Dombroski, Beth A. , Cheng, Po‐Liang , Si, Yaqin , Tucci, Albert , Patil, Vishakha, Valiente‐Banuet, Leopoldo , Li, Chong , Farrell, Kurt , Molina‐Porcel, Laura , Rajput, Alex , Gearing, Marla , Newell, Kathy L. , Rábano‐Gutierrez, Alberto, Oertel, Wolfgang H. , Respondek, Gesine , Stamelou, María , Arzberger, Thomas , Roeber, Sigrun, Müller, Ulrich , Pástor, Pau , Ber, Isabelle Le, Beach, Thomas G. , Litvan, Irene , Rademakers, Rosa , Ross, Owen A. , Boxer, Adam L. , Seeley, Willian W., Lee, Edward B. , Silva, Rohan de , Crary, John F. , Goate, Alison , Friedman, Jeffrey S., Compta, Yaroslau , Coppola, Giovanni , Wang, Li‐San , Group, PSP Genetics Study, Tzeng, Jung‐Ying , Geschwind, Daniel H. , Schellenberg, Gerard D. , Si, Ya-Qin , Mclean, Catriona , Molina-Porcel, Laura , Rajput, Alex , Deyn, Peter Paul De , Le Bastard, Nathalie , Gearing, Marla , Donker Kaat, Laura , Van Swieten, John C , Ghetti, Bernardino F , Troakes, Claire , de Yébenes, Justo G , Meller, Tina , Oertel, Wolfgang H , Respondek, Gesine , Hopfner, Franziska , Pastor, Pau , Brice, Alexis , Dürr, Alexandra , Beach, Thomas G , Serrano, Geidy E , Hazrati, Lili‐Naz , Litvan, Irene , Ross, Owen A , Galasko, Douglas , Boxer, Adam , Miller, Bruce L , Deerlin, Vivianna M. Van , Lee, Edward B , White, Charles L , Morris, Huw R , Crary, John F , Goate, Alison M , Leung, Yuk Yee , Wang, Li-San , Dickson, Dennis W , Höglinger, Günter U , Tzeng, Jung-Ying , Schellenberg, Gerard D , Lee, Wan-Ping , Dombroski, Beth , Farrell, Kurt , McLean, Catriona , Kaat, Laura Donker , Swieten, John C. van , Dopper, Elise G.P. , Ghetti, Bernardino , Troakes, Claire , Müller, Ulrich, Hopfner, Franziska , Brice, Alexis , Ber, Isabelle Le , Serrano, Geidy E. , Galasko, Douglas , Miller, Bruce L. , White, Charles L. , Morris, Huw R. , Naj, Adam Christian , Group, PSP Genetics Study, Dalgard, Clifton L. , Dickson, Dennis W. , Höglinger, Günter U.

Mayo Clinic

Movement Disorders | March 12, 2025 | Volume: 40 | Issue: 5 | Pages: 950-961

10.1002/mds.30150 | PMID 40055946 Get It

AbstractBackgroundThe 17q21.31 region with various structural forms characterized by the H1/H2 haplotypes and three large copy number variations (CNVs) represents the strongest risk locus in progressive supranuclear palsy (PSP).ObjectiveTo investigate the association between CNVs and structural forms on 17q.21.31 with the risk of PSP.MethodsUtilizing whole genome sequencing data from 1684 PSP cases and 2392 controls, the three large CNVs (α, β, and γ) and structural forms within 17q21.31 were identified and analyzed for their association with PSP.ResultsWe found that the copy number of γ was associated with increased PSP risk (odds ratio [OR] = 1.10, P = 0.0018). From H1β1γ1 (OR = 1.21) and H1β2γ1 (OR = 1.24) to H1β1γ4 (OR = 1.57), structural forms of H1 with additional copies of γ displayed a higher risk for PSP. The frequency of the risk sub‐haplotype H1c rises from 1% in individuals with two γ copies to 88% in those with eight copies. Additionally, γ duplication up‐regulates expression of ARL17B, LRRC37A/LRRC37A2, and NSFP1, while down‐regulating KANSL1. Single‐nucleus RNA‐seq of the dorsolateral prefrontal cortex analysis reveals γ duplication primarily up‐regulates LRRC37A/LRRC37A2 in neuronal cells.ConclusionsThe copy number of γ is associated with the risk of PSP after adjusting for H1/H2, indicating that the complex structure at 17q21.31 is an important consideration when evaluating the genetic risk of PSP. © 2025 The Author(s). Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.
A systematic review and meta-analysis of the response to placebo in clinical trials of inclusion body myositis
Naddaf, Elie , Dimachkie, Mazen M. , Benveniste, Olivier, Mandrekar, Jay , Murad, Mohammed , Naddaf, Elie , Dimachkie, Mazen M , Skolka, Michael P , Prokop, Larry J. , Hogrel, Jean‐Yves , Benvéniste, Olivier , Wang, Zhen , West, Colin P.

Mayo Clinic

Rheumatology | March 11, 2025 | Volume: 64 | Issue: 7 | Pages: 4125-4132

10.1093/rheumatology/keaf146 | PMID 40085004 Get It

Abstract Objectives IBM is characterized by slowly progressive muscle weakness making it challenging to detect significant changes in weakness during a clinical trial. Trial participants receiving placebo may behave differently from in natural history studies. We aimed to quantify the change in muscle strength and IBM Functional Rating Scale (IBMFRS) of IBM patients receiving placebo during clinical trials. Methods Several databases were searched. Randomized, double-blinded, placebo-controlled trials without treatment intervention on the placebo group were included. Standardized mean differences (SMD) for change in muscle strength and mean differences for IBMFRS were used to calculate pooled effects, using DerSimonian-Laird continuous random effect models. Meta-regression determined change in muscle strength. Heterogeneity was evaluated using the I2 indicator. Results Eleven eligible trials were identified with 257 participants receiving placebo and a low risk of bias per the Cochrane Collaboration Risk of Bias 2 (RoB2) tool. Participants receiving placebo on average had a measurable decline in muscle strength with a mean effect size of –0.398 (95% CI –0.652, –0.144) (P = 0.002). SMD changed by –0.009 (–0.016, –0.002) points per week (P = 0.015). Manual muscle testing was associated with higher heterogeneity (I2 = 67.68%) compared with quantitative muscle testing (I2 = 0%). Only three studies reported IBMFRS results. The pooled change in IBMFRS at 12 months was –2.189 points (–3.893, –0.485) on this 0 to 40-point scale (P = 0.012), with relatively high heterogeneity. Conclusion Participants with IBM displayed a measurable decline in their muscle strength and IBMFRS during clinical trials, in keeping with the disease’s slowly progressive nature. These estimates can inform sample size calculations in future studies.
Analysis of the BraTS 2023 Intracranial Meningioma Segmentation Challenge
LaBella, Dominic , Baid, Ujjwal , Khanna, Omaditya , McBurney-Lin, Shan , McLean, Ryan, Rashid, Arif , Altes, Talissa A. , Bhalerao, Radhika , Dhemesh, Yaseen , Hilal, Fathi , Janas, Anastasia , Kazerooni, Anahita Fathi , Kent, Collin , Leu, Kevin , Maleki, Nazanin , Pajot, Maxence , Saluja, Rachit , Wang, Chunhao , Warman, Pranav , Adewole, Maruf , Albrecht, Jake, Anwar, Syed Muhammad , Bergquist, Timothy , Chen, Sully Francis, Conte, Gian-Marco, Eddy, J. Mark , Ezhov, Ivan , Jiang, Zhifan , Johanson, Elaine, Linguraru, Marius George , Liu, Xinyang , Meier, Zeke, Shinohara, Russell Takeshi, Abayazeed, Aly, Akinola, Rachel , Bilello, Michel , Verdier, Maria Correia de, Moassefi, Mana, Pease, Matthew W., Vollmuth, Phillipp, Diffley, David, Hussain, Ali , Fehringer, Pascal, Bronstein, Yulia, Deptula, Lisa, Stein, Evan G. , Taherzadeh, Mahsa , Oliveira, Eduardo Portela de, Haughey, Aoife , Kontzialis, Marinos , Saba, Luca , Turner, Benjamin , Brüßeler, Melanie, Gkampenis, Athanasios , Weiss, David Maximilian, Mansour, Aya, Shawali, Islam H., Yordanov, Nikolay , Hourani, Roula , Moshebah, Mohammed Yahya, Abouelatta, Ahmed Magdy, Willms, Klara, Martin, Dann C., Okar, Abdullah , Kite, Dominic , Haider, Muhammad Ammar , Aristizábal, Alejandro , Karargyris, Alexandros , Kassem, Hasan , Sheller, Micah , Alonso‐Basanta, Michelle , McLean, Ryan, Nedelec, Pierre , Rashid, Arif, Tahon, Nourel Hoda , Godfrey, Devon , Floyd, Scott , Kirkpatrick, John P. , Kofler, Florian , Menze, Bjoern H. , Reitman, Zachary J. , Rudie, Jeffrey D. , Velichko, Yury , Anazodo, Udunna , Chung, Verena , Chai, Rong, Dako, Farouk , Eddy, James, Khalili, Nastaran , Iglesias, Juan Eugenio , Leemput, Koen Van , Li, Hongwei , Mahtabfar, Aria , Moawad, Ahmed W. , Mongan, John , Piraud, Marie , Shinohara, Russell Taki , Wiggins, Walter F. , Jakab, András Áron , Verdier, Maria Correia de , Crivellaro, Priscila Sacilotto , Davatzikos, Christos , Farahani, Keyvan , Freymann, John , Hess, Christopher Paul , Huang, Raymond Y. , Lohmann, Philipp , Moassefi, Mana , Sollmann, Nico , Nandolia, Khanak , Warren, Daniel I, Saba, Luca , Ansari, Shehbaz , Stein, Joel M. , Rizvi, Tanvir , D’Anna, Gennaro , Taha, Ahmed , Sharifi, Yasaman , Faghani, Shahriar , Pinho, Marco C. , Pati, Sarthak , Villanueva-Meyer, Javier E. , Rauschecker, Andreas M. , Nada, Ayman , Aboian, Mariam , Flanders, Adam E , Wiestler, Benedikt , Bakas, Spyridon , Calabrese, Evan

Mayo Clinic

Machine Learning for Biomedical Imaging | March 7, 2025 | Volume: 3 | Issue: March 2025 | Pages: 38-58

10.59275/j.melba.2025-bea1 Get It

We describe the design and results from the BraTS 2023 Intracranial Meningioma Segmentation Challenge. The BraTS Meningioma Challenge differed from prior BraTS Glioma challenges in that it focused on meningiomas, which are typically benign extra-axial tumors with diverse radiologic and anatomical presentation and a propensity for multiplicity. Nine participating teams each developed deep-learning automated segmentation models using image data from the largest multi-institutional systematically expert annotated multilabel multi-sequence meningioma MRI dataset to date, which included 1000 training set cases, 141 validation set cases, and 283 hidden test set cases. Each case included T2, FLAIR, T1, and T1Gd brain MRI sequences with associated tumor compartment labels delineating enhancing tumor, non-enhancing tumor, and surrounding non-enhancing FLAIR hyperintensity. Participant automated segmentation models were evaluated and ranked based on a scoring system evaluating lesion-wise metrics including dice similarity coefficient (DSC) and 95% Hausdorff Distance. The top ranked team had a lesion-wise median dice similarity coefficient (DSC) of 0.976, 0.976, and 0.964 for enhancing tumor, tumor core, and whole tumor, respectively and a corresponding average DSC of 0.899, 0.904, and 0.871, respectively. These results serve as state-of-the-art benchmarks for future pre-operative meningioma automated segmentation algorithms. Additionally, we found that 1286 of 1424 cases (90.3%) had at least 1 compartment voxel abutting the edge of the skull-stripped image edge, which requires further investigation into optimal pre-processing face anonymization steps.
Time-restricted eating and autosomal dominant polycystic kidney disease: a pilot, randomized clinical trial
Steele, Cortney , Ostrow, Anna, Coleman, Erin R., George, Diana, Bing, Kristen , Ramanathan, Sumana, Gregory, Adriana , Kline, Timothy L. , Thomas, Elizabeth A. , Chonchol, Michel , Nowak, Kristen L. , Nowak, Kristen L , Wang, Wei Qiang , Gitomer, Berenice Y , Thomas, Elizabeth

Mayo Clinic

Clinical Kidney Journal | March 5, 2025 | Volume: 18 | Issue: 4 | Pages: No page available

10.1093/ckj/sfaf069 | PMID 40207097 Get It

ABSTRACT Background Autosomal dominant polycystic kidney disease (ADPKD) is the most commonly inherited progressive kidney disease. Time-restricted eating (TRE) is a fasting regimen that restricts eating to a particular window (typically 8 hours/day), which could slow cyst growth based on preclinical models. Methods A 12-month, randomized, controlled, behavioral dietary intervention compared TRE with a control group given healthy eating advice without TRE (HE), without caloric restriction. Participants underwent baseline and 12-month measurements, including adherence (percentage of participants adhering to the 8-hour window; primary outcome), and MRI to determine height-adjusted total kidney volume (htTKV) and adiposity. Results Twenty-nine participants (23 females, mean standard ± deviation 48 ± 9 years) with a body mass index of 31.1±5 kg/m2 were randomized to TRE (n = 14) or HE (n = 15). Of the total participants, 71% (n = 10) of TRE and 87% (n = 13) of HE participants completed the intervention. The eating window was 9.6 ± 3.6 hours for TRE (60% achieving the 8-hour window) and 12.0 ± 2.0 for HE groups (P = .07). At month 12, both groups lost modest weight (−2.4 ± 6.4% and −3.6 ± 5.4% in the TRE and HE groups, respectively). Annual change in htTKV was 3.0 ± 8.5% and 4.6 ± 8.8% in the TRE and HE groups, respectively. Both change in weight (r = 0.67, P < .01) and change in visceral adiposity (r = 0.54, P < .01) were positively correlated with change in htTKV. Conclusion Both the TRE and HE group lost modest weight at 12 months. The targeted TRE adherence of ≥75% of participants was not achieved. Weight and adiposity loss may be more important drivers of kidney growth than the timing of eating.
Intensity-Based Assessment of Hippocampal Segmentation Algorithms Using Paired Precontrast and Postcontrast MRI
Cramer, Justin , Baxter, Leslie C. , Lang, Harrison, Parker, Jonathon J. , Matthees, Nicholas , Ikuta, Ichiro , Wang, Yalin , Parker, Jonathon, Chen, Alicia , Matthees, Nicholas , Ikuta, Ichiro , Wang, Yalin , Baxter, Leslie C. , Lang, Harrison, Chen, Alicia M. , Zhou, Yuxiang

Mayo Clinic

Bioengineering | March 4, 2025 | Volume: 12 | Issue: 3 | Pages: 258

10.3390/bioengineering12030258 | PMID 40150721 Get It

Hippocampal segmentation is essential in neuroimaging for evaluating conditions like Alzheimer’s dementia and mesial temporal sclerosis, where small volume changes can significantly impact normative percentiles. However, inaccurate segmentation is common due to the inclusion of non-hippocampal structures such as choroid plexus and cerebrospinal fluid (CSF), leading to volumetric overestimation and confounding of functional analyses. Current methods of assessment largely rely on virtual or manual ground truth labels, which can fail to capture these inaccuracies. To address this shortcoming, this study introduces a more direct voxel intensity-based method of segmentation assessment. Using paired precontrast and postcontrast T1-weighted MRIs, hippocampal segmentations were refined by adding marginal gray matter and removing marginal CSF and enhancement to determine a total required correction volume. Six segmentation algorithms—e2dhipseg, HippMapp3r, hippodeep, AssemblyNet, FastSurfer, and QuickNat—were implemented and compared. HippMapp3r and e2dhipseg, followed closely by hippodeep, exhibited the least total correction volumes, indicating superior accuracy. Dedicated hippocampal segmentation algorithms outperformed whole-brain methods.
Urinary control in cervical myelopathy: does it improve postsurgery? A Quality Outcomes Database study
Shabani, Saman , Le, Vivian , Jamieson, Alysha, Chan, Andrew K. , Johnson, Sarah E. , Wang, Michael Y. , Haid, Regis W., Knightly, John J. , Gottfried, Oren N., Shaffrey, Christopher I. , Virk, Michael S. , Shaffrey, Mark E., Foley, Kevin T. , Coric, Domagoj , Upadhyaya, Cheerag D., Potts, Eric A. , Turner, Jay D. , Tumialán, Luis M. , Chou, Dean , Fu, Kai-Ming, Asher, Anthony L. , Bisson, Erica F. , Bydon, Mohamad , Mummaneni, Praveen V. , Lavadi, Raj Swaroop , Agarwal, Nitin , Letchuman, Vijay , Park, Paul , Uribe, Juan S.

Mayo Clinic

Journal of Neurosurgery: Spine | February 28, 2025 | Volume: 42 | Issue: 5 | Pages: 541-550

10.3171/2024.10.spine24338 | PMID 40020227 Get It

OBJECTIVE Cervical spondylotic myelopathy (CSM) can be associated with urinary dysfunction, leading to an inability or marked difficulty with micturition. This study aimed to evaluate the urinary dysfunction, long-term prognosis, and recovery in patients with CSM following surgical intervention. METHODS The CSM cases of the Quality Outcomes Database SpineCORe study group were analyzed. Urinary control was assessed using the modified Japanese Orthopaedic Association (mJOA) urinary function subscore. Improvement was defined as a minimum improvement of 1 point in the mJOA urinary function subscore at the 2-year follow-up. Univariate and multivariable analyses were conducted as appropriate. RESULTS Of 1141 patients, 772 (67.7%) patients were identified with a minimum 2-year follow-up mJOA score, and 249 (32.3%) of these patients reported baseline urinary dysfunction. Of those 249 patients with baseline urinary control problems, 193 (77.5%) had improvement in urinary function postoperatively, and more women than men had improved urinary control after CSM surgery (54.9% vs 45.1%, p = 0.03). Apart from sex, demographic characteristics of patients who experienced urinary function improvement versus those who did not were similar. Patients who experienced urinary function improvement had lower overall baseline mJOA scores (10.2 vs 13, p < 0.01). Both cohorts reported similar rates of postoperative satisfaction (North American Spine Society scores of 1 and 2) (89.1% vs 85.9%, p = 0.26). CONCLUSIONS Among the one-third of patients with CSM experiencing urinary dysfunction, nearly 78% achieve improved urinary function at the 2-year postoperative follow-up. Women are more likely to experience improvement in this function. Patients with and without improvements in urinary function are equally satisfied with their surgery and report similar outcomes on long-term follow-up.
Chemogenetic activation of microglial Gi signaling decreases microglial surveillance and impairs neuronal synchronization
Wang, Lingxiao , Kleidonas, Dimitrios , Qi, Fangfang , Liang, Yue , Zheng, Jiaying , Wu, Long‐Jun , Zhao, Shunyi , Wang, Lingxiao , Kleidonas, Dimitrios , Qi, Fangfang , Liang, Yue , Zheng, Jiaying , Umpierre, Anthony D , Wu, Long-Jun , Zhao, Shunyi , Umpierre, Anthony D.

Mayo Clinic

Science Advances | February 28, 2025 | Volume: 11 | Issue: 9 | Pages: No page available

10.1126/sciadv.ado7829 | PMID 40020068 Get It

Microglia actively survey the brain and dynamically interact with neurons to maintain brain homeostasis. Microglial Gi protein–coupled receptors (Gi-GPCRs) play a critical role in microglia-neuron communications. However, the impact of temporally activating microglial Gi signaling on microglial dynamics and neuronal activity in the homeostatic brain remains largely unknown. In this study, we used Gi-based designer receptors exclusively activated by designer drugs (Gi-DREADD) to selectively and temporally modulate microglial Gi signaling pathway. By integrating this chemogenetic approach with in vivo two-photon imaging, we observed that exogenous activation of microglial Gi signaling transiently inhibited microglial process dynamics, reduced neuronal activity, and impaired neuronal synchronization. These altered neuronal functions were associated with a decrease in interactions between microglia and neuron somata. Together, this study demonstrates that acute, exogenous activation of microglial Gi signaling regulates neuronal circuit function, offering a potential pharmacological target for the neuromodulation through microglia.
SensitiveCancerGPT: Leveraging Generative Large Language Model on Structured Omics Data to Optimize Drug Sensitivity Prediction
Chowdhury, Shaika , Wang, Liewei , Yang, Ping , Zong, Nansu , Rajaganapathy, Sivaraman , Sun, Lichao, Sun, Lichao , Cerhan, James R.

Mayo Clinic

No container title available | February 28, 2025 | Volume: No volume available | Issue: No issue available | Pages: No page available

10.1101/2025.02.27.640661 | PMID 40060567 Get It

Abstract Objective The fast accumulation of vast pharmacogenomics data of cancer cell lines provide unprecedented opportunities for drug sensitivity prediction (DSP), a crucial prerequisite for the advancement of precision oncology. Recently, Generative Large Language Models (LLM) have demonstrated performance and generalization prowess across diverse tasks in the field of natural language processing (NLP). However, the structured format of the pharmacogenomics data poses challenge for the utility of LLM in DSP. Therefore, the objective of this study is multi-fold: to adapt prompt engineering for structured pharmacogenomics data toward optimizing LLM’s DSP performance, to evaluate LLM’s generalization in real-world DSP scenarios, and to compare LLM’s DSP performance against that of state-of-the-science baselines. Methods We systematically investigated the capability of the Generative Pre-trained Transformer (GPT) as a DSP model on four publicly available benchmark pharmacogenomics datasets, which are stratified by five cancer tissue types of cell lines and encompass both oncology and non-oncology drugs. Essentially, the predictive landscape of GPT is assessed for effectiveness on the DSP task via four learning paradigms: zero-shot learning, few-shot learning, fine-tuning and clustering pretrained embeddings. To facilitate GPT in seamlessly processing the structured pharmacogenomics data, domain-specific novel prompt engineering is employed by implementing three prompt templates (i.e., Instruction, Instruction-Prefix, Cloze) and integrating pharmacogenomics-related features into the prompt. We validated GPT’s performance in diverse real-world DSP scenarios: cross-tissue generalization, blind tests, and analyses of drug-pathway associations and top sensitive/resistant cell lines. Furthermore, we conducted a comparative evaluation of GPT against multiple Transformer-based pretrained models and existing DSP baselines. Results Extensive experiments on the pharmacogenomics datasets across the five tissue cohorts demonstrate that fine-tuning GPT yields the best DSP performance (28% F1 increase, p-value= 0.0003) followed by clustering pretrained GPT embeddings (26% F1 increase, p-value= 0.0005), outperforming GPT in-context learning (i.e., few-shot). However, GPT in the zero-shot setting had a big F1 gap, resulting in the worst performance. Within the scope of prompt engineering, performance enhancement was achieved by directly instructing GPT about the DSP task and resorting to a concise context format (i.e., instruction-prefix), leading to F1 performance gain of 22% (p-value=0.02); while incorporation of drug-cell line prompt context derived from genomics and/or molecular features further boosted F1 score by 2%. Compared to state-of-the-science DSP baselines, GPT significantly asserted superior mean F1 performance (16% gain, p-value<0.05) on the GDSC dataset. In the crosstissue analysis, GPT showcased comparable generalizability to the within-tissue performances on the GDSC and PRISM datasets, while statistically significant F1 performance improvements on the CCLE (8%, p-value=0.001) and DrugComb (19%, p-value=0.009) datasets. Evaluation on the challenging blind tests suggests GPT’s competitiveness on the CCLE and DrugComb datasets compared to random splitting. Furthermore, analyses of the drug-pathway associations and log probabilities provided valuable insights that align with previous DSP findings. Conclusion The diverse experiment setups and in-depth analysis underscore the importance of generative LLM, such as GPT, as a viable in silico approach to guide precision oncology. Availability https://github.com/bioIKEA/SensitiveCancerGPT