Success story: Novartis

It’s no secret that drug development is a time-intensive, costly process. It takes an initial 5,000 to 10,000 compounds to obtain a single approved drug, a winnowing and analysis process that can take 10 to 15 years and cost approximately $1 billion [Source: Pharmaceutical Researchers and Manufacturers Association]. Every stage – from pre-discovery, where drug targets are identified, to drug discovery, where candidate drugs are tested, to preclinical laboratory and animal testing – generates volumes of specimens requiring pathological analysis. And since it is always preferable to eliminate poor candidates as early as possible, time is of the essence.

At Novartis Institutes for Biomedical Research (NIBRI), pathologists were challenged to provide timely analysis with quantifiable measurements for the many thousands of samples the company generates in both early research and later preclinical development phases. Pathologists analyzed biomarkers and microscopic tissue parameters to understand efficacies, toxicity and mechanism of action; these insights provided critical guidance to the drug discovery effort.

While handling the massive volumes of material with improved turnaround time was a top goal for researchers, of equal importance was the ability to deliver results that were not subjectively developed, but were instead quantifiably measured. This was particularly important given the large volume. Analysis needed to be consistently applied so that results could be reliably aggregated.

Recognizing that reliance on manual analysis of glass slides was not practical, Novartis turned to digital pathology. By digitizing the slide, the company could easily use image analysis algorithms to quantifiably measure effects. The process introduced automation into previously tedious and labor-intensive processes, and helped Novartis pathologists handle the voluminous workload generated by drug research and discovery, and shorten their turnaround time.

Solution

Novartis installed two high capacity ScanScopes from Aperio for routine histology scanning, and also purchased Aperio’s image analysis package. The systems support five different departments at the NIBRI.
In their first two years at Novartis, ScanScopes scanned over 20,000 slides, nearly all of which were subjected to image analysis. Slides were scanned for a variety of uses, but the dominant use was high volume pre-screens of IHC-staining on paraffin embedded specimens.

Novartis found it most frequently used the positive pixel count algorithm, a multipurpose algorithm that measures area and intensity of staining for two-colored slides. Also useful were the nuclear algorithm for counting the percentage of nuclei that stained positive, and the color deconvolution algorithm that accurately separates the image into three channels so that each stain can be separately measured, even when the stains are superimposed. “We have found the application of algorithms to digitally scanned slides to be a very convenient way of quantitating change,” explained Frank Voelker, former head of the Tissue Biomarker Laboratory at Novartis Institutes of Biomedical Research. “It’s much better than the old way, where samples were graded subjectively.” The classical method of evaluating tumors is an H-score that estimates an average measurement of staining. H-scores are problematic because they fail to distinguish between, for instance, a high percentage of medium cells and a small percentage of strong cells. Moreover, the process is time-consuming and tedious, and is highly subjective. NIBRI’s comparisons of H-scores to image analysis-derived scores show they track and are an effective measure. It is the organization’s hope to supplement or replace subjective H-scores with quantitative image analysis scores such as average cytoplasmic intensity as a standard form of measurement.

Benefits/value

The benefits of the Aperio systems were immediately recognized, as researchers saw an over 30-fold improvement in turnaround time. The systems’ automated nature and entire-slide approach made for a streamlined workflow that enabled researchers to focus on analyzing the results, rather than on taking many tedious steps to obtain results. Faster turnaround time ties directly to research efficiency and costs, making it easy for Novartis to see an immediate return on its ScanScope system investment.

Also beneficial to researchers was the fact that algorithms provided a means to quantitate effects and eliminate subjective estimates or counts to better the drug research development process. Results could be provided with more confidence in the underlying methodology used to achieve them, and could be more readily compared across different projects, for additional research impact.

Next steps

NIBRI continues to support a broad range of early development projects, but is also in the process of developing practical procedures and methods appropriate for the analysis of oncology specimens, in support of later-development projects. It continues to hone its use of algorithms, so that it can expand its sphere of automation and quantifiable analysis, and has been working more extensively with Aperio’s color deconvolution algorithm, for instances when slides have received a multi-stain.
The company is also working to take better advantage of Spectrum to improve its digital pathology management, access, and organization capabilities.

About Aperio Technologies, Inc.

Aperio provides systems and services for digital pathology. Our award-winning ScanScope® slide scanning systems and Spectrum™ digital pathology information management software are helping pathology professionals around the world improve the efficiency and quality of pathology services and improve the bottom line. Applications include remote viewing, archival and retrieval, digital conferencing, research and education, and image analysis. For more information and to view a gallery of digital slides created by ScanScope Systems, visit www.aperio.com.