Duodenal adenoma organoids

Innovations in Multimodal Image Analysis Help Improve Biopharma Productivity

While a productivity crisis in the pharmaceutical industry still exists, we have seen some promising improvement thanks to innovations in technology, the use of human-relevant disease models, combined with ‘omics technologies and advanced imaging techniques. These developments are providing deeper scientific insights into disease mechanisms, improving target selection, and refining validation methods.

The Biopharma Productivity Challenge

In the early 2000s, the pharmaceutical industry faced a significant productivity crisis^(1-3). Drug development pipelines stagnated as attrition rates in clinical trials soared. Costs escalated while the time to bring new drugs to market increased. A 2009 analysis published in Nature Reviews Drug Discovery concluded⁽⁴⁾, “nothing that companies have done in the past 60 years has affected their rates of new-drug production,” including mergers, acquisitions, reorganizations, and process improvement.

By the 2020s, however, there was a surprising shift. Attrition rates started to decline, and the number of novel drugs approved annually began to climb⁽⁵⁾. In 2022, US Food and Drug Administration (FDA) approvals for new drugs were at their second-highest level measured by 2022, surpassed only by 2018^(6,7). This uptick has been attributed to a confluence of factors⁽⁵⁾. Advances in genomics have reached a threshold where insights can be directly applied to actionable results, such as phenotyping diseases and identifying oncology biomarkers. Artificial intelligence (AI), 3D printing, and automation are streamlining research processes and enhancing decision-making. Novel therapies like CAR T-cell and CRISPR-based gene editing are opening new frontiers in treatment.

Among these transformative developments, advances in imaging technology and multimodal analysis stand out as pivotal contributors to the acceleration of drug discovery. High-throughput screening (HTS) technologies and sophisticated imaging methods have allowed researchers to rapidly evaluate complex biological models at the phenotypic level, eliminating unpromising drug mechanisms earlier in the pipeline. These innovations are not only accelerating drug development timelines but are also reducing costs, enabling the industry to produce more effective treatments.

The Role of Multimodal Image Analysis in Drug Discovery

Advances in multimodal imaging techniques, particularly HTS methods⁽⁸⁾, are revolutionizing the drug discovery landscape. One notable innovation is cell painting, a high-content, image-based assay that uses fluorescent dyes to create rich morphological profiles of cells (Figure 1). This technique enables researchers to analyze cellular states at both morphological and molecular levels, offering profound insights into disease biology.

Predictive Applications of Cell Painting

Cell Painting assay

Figure 1: Cell Painting assay. Cells were compound treated, stained, then imaged using the ImageXpress Micro Confocal system. Example images of each acquired channel from a control well is shown. The last panel shows a composite image consisting of actin, endoplasmic reticulum (ER), and nuclei staining⁽⁹⁾.

Cell painting can be combined with transcriptomic data to predict how various perturbations affect cellular function. Studies by the Broad Institute of MIT and Harvard, such as the research led by Anne Carpenter and Shantanu Singh⁽¹⁰⁾, have shown how integrating morphological profiling with transcriptomics provides overlapping yet distinct insights into cellular states. This dual approach enhances our ability to decode how different therapeutic perturbant influence cellular gene expression.

Cell painting is being used to study compound libraries, providing valuable information about compound diversity and identifying potential leads more efficiently. In cancer research, this technique has proven invaluable for understanding drug resistance mechanisms^(11,12). By profiling how cells respond morphologically to specific drugs, researchers can identify mechanisms likely to induce resistance in cancer cells, allowing them to refine drug development strategies early in the pipeline.

Identifying Disease Signatures and Genetic Functions

Cell painting also excels at identifying phenotypic disease signatures and uncovering the functional impacts of genetic variations. Recent studies highlight its ability to predict the functional impact of somatic variants associated with diseases like lung cancer⁽¹³⁾. Using techniques like cell morphology variant impact prediction (cmVIP), researchers can determine how genetic variants affect cellular morphology⁽¹³⁾. These findings are instrumental in linking genetic variations to biological mechanisms, helping uncover the roots of disease.

This method is particularly powerful in identifying gene functions and chemical regulators of disease profiles. By systematically analyzing morphological changes induced by genetic and chemical perturbations, researchers can uncover previously uncharacterized genetic contributions to disease and identify potential therapeutic targets⁽¹⁴⁾.

Toxicity and Chemical Profiling

Another critical application of multimodal imaging is assessing drug toxicity⁽¹⁵⁾. By systematically profiling morphological changes induced by genetic and chemical perturbations, researchers can predict adverse effects at an early stage. This fail-faster approach saves resources by eliminating unsafe compounds before they progress into costly clinical trials.

Evolution of Screening Technologies: Enabling Fail-Faster Strategies

The incorporation of new technologies to screening paradigms for target identification and hit identification is accelerating the pace of drug discovery, enabling researchers to evaluate thousands of compounds in parallel. Recent advancements have further enhanced the speed and precision of these new technologies and screening approaches, streamlining the process of identifying viable drug candidates.

CRISPR-Enhanced Phenotypic Screening

The integration of CRISPR technology with cell-based screening has opened new avenues to study drug-target interactions^(16,17). By modifying specific genomic targets, researchers can observe how small molecules affect these sites, leading to more accurate identification of potential drug candidates. This approach allows for a more precise understanding of the biological pathways involved, helping researchers prioritize the most promising compounds while discarding less viable candidates, thus reducing the time spent on unviable compounds.

Automated Proteomics Platforms

Automated proteomics platforms like autoSISPROT⁽¹⁸⁾ are transforming target identification by enabling the rapid identification of drug targets and off-targets. Processing up to 96 samples in under 2.5 hours, these platforms enhance throughput tenfold compared with traditional methods. Their speed and precision allow researchers to validate targets quickly.

AI-Assisted Hit Prioritization

Artificial intelligence is reshaping workflows through machine learning algorithms that prioritize drug candidates. By filtering out false positives, these systems ensure that resources are allocated to truly bioactive compounds. This data-driven approach enhances reproducibility and reduces costs, making early-stage screening more efficient and reliable.

Complex Cell Models: Bridging the Gap Between Preclinical and Clinical Research

While imaging and HTS technologies are crucial, the use of advanced cell models has been equally transformative in drug discovery. Organoids, three-dimensional cellular models derived from patient tissues (Figure 2), have emerged as powerful tools for studying disease mechanisms and evaluating therapeutic interventions.

Organoids in Disease Characterization and Research

Brain Organoids derived from a patient with Pitt-Hopkins Syndrome

Figure 2: Microscopy images reveal significant differences in size and structure between brain organoids derived from a patient with Pitt-Hopkins Syndrome (right) and from a control (left) [Credit: UC San Diego Health Sciences^(19,21)].

Organoids are self-assembling 3D cellular models derived from patient tissues that mimic the architecture and functionality of organs. They provide a more physiologically relevant platform for studying disease mechanisms than two-dimensional cell culture models. For instance, patient-derived organoids (PDOs)^(19,21) have been used in cancer research to recreate tumor heterogeneity and test personalized therapies, enabling detailed investigations into tumor biology and therapeutic responses^(21,22). By co-culturing PDOs with CAR T-cells, researchers are optimizing immunotherapies for solid tumors, addressing challenges like immune evasion and resistance, to enhance efficacy.

These human-relevant models have also proven valuable in liver cancer research, where they are used to study hepatocellular carcinoma⁽²³⁾ and cholangiocarcinoma (22). By recreating the early stages of cancer progression and testing personalized treatments, organoids bridge the gap between preclinical and clinical research, offering a scalable and reliable tool for drug discovery.

Scalable Production of Organoids

Technologies like automated 3D culture platforms are enabling the scalable production of organoids. These tools are invaluable for rapidly assessing the safety of new compounds and evaluating how well drugs target disease-specific pathways. The ability to generate assay-ready organoids at scale provides researchers with consistent models for higher volume or throughput workflows.

Technological advancements, such as proprietary bioreactors, are making organoids more accessible⁽²⁴⁾. Proprietary bioreactor systems support the rapid and reproducible generation of organoids, making it easier to model complex human tissues for applications like drug toxicity testing and therapeutic efficacy screening. The technology allows researchers to access large quantities of uniform organoids, addressing the challenge of variability often associated with traditional organoid culturing methods. Combined with advanced imaging and analytical solutions tailored to these 3D models, high-content phenotypic profiling is now also rapidly achievable at scale. The integration of precision engineering with biological expertise to offer solutions like 3D Ready Organoids⁽²⁵⁾ empowers researchers to bridge the gap between in vitro and in vivo studies, delivering insights that are both actionable and clinically relevant. This innovation not only streamlines the drug discovery process but also enhances its predictive power, reducing the risks of late-stage failures.

Automation and Artificial Intelligence in Drug Discovery

The integration of AI and automation is accelerating every stage of the drug discovery process, from target identification to lead optimization^(26-28). Automated platforms like the CellXpress.ai automated cell culture system⁽²⁹⁾ streamline the cultivation of cell models, addressing one of the most labor-intensive aspects of drug discovery and ensuring consistency and scalability. By automating routine tasks like media exchange, cell plating, and passaging, the CellXpress.ai cell culture system significantly reduces the potential for human error and ensures reproducibility across experiments. Its advanced AI-driven algorithms monitor cell growth and morphology in real-time, enabling researchers to make informed decisions about culture conditions without manual intervention. This system is particularly well-suited for generating consistent batches of assay-ready cells, which are critical for HTS and other downstream applications. By integrating seamlessly with high-content imaging and analysis tools, these solutions accelerate workflows, enhances scalability, and free up valuable researcher time to focus on experimental design and data interpretation.

Artificial intelligence is playing a pivotal role in toxicology, using machine learning models to predict drug toxicity and adverse mechanisms before they occur. These models leverage vast datasets to identify potential toxicities early, before in vivo testing, reducing the reliance on animal testing and expediting preclinical evaluations. Tools driven by AI, like the DeepChem Python library^(26,30), are revolutionizing drug screening by predicting chemical properties and generating novel compounds. These digital tools enable researchers to design drugs that are not only effective but also have favorable safety profiles, making the drug discovery process more efficient and precise.

Shaping the Future

The pharmaceutical industry’s productivity crisis of the early 2000s appears to be reversing, thanks to several key drivers, including the maturation of genomics, the adoption of advanced technologies like AI and 3D printing, and innovations in imaging, multimodal analysis, and advanced cell models. Techniques like cell painting are providing deeper insights into disease biology, while HTS technologies enable fail-faster approaches that save resources. Meanwhile, AI and automation are ensuring that these advancements are applied efficiently.

As the knowledge base in genomics and molecular biology continues to mature, and as new tools like automated 3D bioprinting become more prevalent, the industry is well-positioned to sustain its momentum in drug discovery and the development of effective treatments. These innovations promise not only to improve bench-to-bedside success rates but also to bring new medicines to patients faster, addressing unmet medical needs and transforming healthcare globally.

References