NEW YORK (GenomeWeb) – This week, SAP unveiled two new solutions for customers in the personalized medicine space: SAP Foundation for Health and SAP Medical Research Insights, both of which are based on SAP Hana, the company's proprietary in-memory database computing technology.
The company discussed both solutions at an SAP Spotlight Tour held in New York. The first of these, SAP Foundation for Health provides flexible and extensible compute infrastructure for integrating, managing, and analyzing structured and unstructured datasets including genomic and electronic medical records data in both research and clinical contexts. The platform features a common database schema, functional libraries, computational models, and algorithms that can be used to directly to mine and query information or customized to fit specific research and clinical needs.
Key components in this first iteration of the software include a clinical data model that enables users to comb clinical data from different electronic health record systems, and a genomic data model that lets users transform information from variant call files into a tabular format designed to enable efficient querying, Enakshi Singh, a senior product manager for genomics and healthcare at SAP, told GenomeWeb at the event. Because the software is based on the SAP Hana in-memory database computing platform, "we have the in-memory benefits as well as ... the data models that have been built to enable the integration and analytics of this data," Singh said.
SAP offers multiple options for accessing and using Foundation for Health. On one hand, customers can use the platform as it stands to tackle their research questions — an option that will likely appeal to more technically inclined users — or they can combine specific components into bespoke applications that let them query and mine data in a more tailored manner working in collaboration with SAP or on their own, she said.
A concrete example of what this looks like in practice is the company's second new product, SAP Medical Research Insights, which is based on Foundation for Health and designed for patient cohort analysis. Medical Research Insights was initially developed as part of a collaboration between SAP and researchers at the National Center for Tumor Disease in Heidelberg, Germany. NCT researchers currently use the software to match patients to clinical trials, among other use cases.
Singh demonstrated the software for attendees at the NY meeting. The software provides tools for filtering and grouping patients based on various user-selected attributes such as type of cancer, tumor histology, and differences in survival rate. Users can also compare similarities and differences between individual patient genomes or compare groups of patient genomes to explore mutations that are correlated with things like survival rates post-treatment, for example.
"With SAP Medical Research Insights, medical staff can visualize and analyze patient data in real time," Christof von Kalle, NCT's head of translational oncology and managing director, said in a statement. "[It] provides a holistic view of a patient's medical history in a graphical timeline, allowing physicians more time to care for patients instead of going through paperwork [and] it also dramatically reduces the turnaround time for clinical studies," he said.
Foundation for Health can be run on premises or on a private cloud hosted by SAP, Singh said. The company has a complex pricing system for its infrastructure that varies depending on factors such as system size, associated services, and in some cases the number of users. For example, according to the company's website, it offers four categories of packages for the cloud version of the Hana platform with multiple purchase options included in each category. Generally speaking, price points range from a free trial for a starter package all the way up to a premium enterprise package that has a list price of $4,064 per month. There's also an option to craft a custom system, and pricing here depends on exactly what's included in the system. In addition the company offers database administrative packages that start at $635 per month and go up to $1,099 per month. In addition to the cost of Hana, there are separate costs for licenses to Foundation for Health and for Medical Research Insights that are based on the number of users.
SAP could eventually commercialize other bespoke applications developed with customers and release them either as separate applications like Medical Research Insights or make them available as components of the Foundation for Health platform, Singh told GenomeWeb.
"We pick very specific projects to help understand the needs and [then] create an application that's tailored to that use case or that component into the Foundation for Health." She added that the company is interested in more collaborations with both industry and academic partners.
It's these partnerships that have driven SAP's product development in the healthcare and biomedicine space, David Delaney, chief medical officer for SAP's healthcare sector, told GenomeWeb. The products showcased this week "really represent the distillation of what we learned working with really amazing partners," he said. "What you'll see over the next quarters are additional use cases ... [and] follow-on work with our current partners."
Partnerships give the company insight into the unique needs of the markets it's interested in and enable it develop tailored offerings that address those needs. For example, there's a lot of unstructured free text generated in the context of oncology. "We've been leveraging SAP's core national language processing capabilities but then adding on healthcare-specific capabilities with linguistic analysis, ontologies, and a data model to create value there," Delaney said.
Moreover, SAP can offer additional services alongside these tailored solutions that customers would not get from standard bioinformatics firms, Delaney added. "There are other things to be really successful at this that you need to do that boutique firms can't do." For example, SAP has a tool that's used by stock markets to process large quantities of data such as business transactions in real time and identify exceptions, which could be brought to bear on large-scale genomic and health datasets. Add to that the speed and agility of the Hana infrastructure and "[it's] a combination that's very difficult for most organizations to counter," he said.
Target customers for Foundation for Health include pharmaceutical and biotechnology companies, academic institutions, contract research organizations, insurance companies, and government entities.
Elite athletes, pathogenic mutations
One of the projects at the meeting offered as an example of Hana's potential for genomics is Stanford University's Exercise at the Limit – Inherited Traits of Endurance (ELITE) study, which aims to characterize genetic determinants of human performance by studying clinical and genomic data from elite athletes defined as individuals with a maximum rate of oxygen consumption of over 80 ml/kg/min. The project is run by Euan Ashley, an associate professor of cardiovascular medicine and genetics, at Stanford.
The ELITE researchers are using Foundation for Health to analyze the data that they are collecting for the study. That includes EMR data, whole-exome and possibly some whole-genome sequencing data, as well as data from wearable devices. In discussing the study at the SAP event, Carlos Bustamante, a population geneticist at Stanford University and one of the investigators on the ELITE study, said that understanding the underlying genetics of these super-performers could have important implications for ongoing efforts to improve human health and extend life.
"It's an exploratory question. Do elite athletes have genetic changes that make them stand out relative to the rest of the population?" he told GenomeWeb at the meeting. "If you think about other traits like human longevity, nobody has found that yet. You sequence centenarians and super-centenarians and they don't really look that different than you and I ... but when it comes to elite athletes, there really is something physiologically different [in] some subset of elite athletes."
The researchers hope to collect data on 400 to 500 'elite' individuals as well as control subjects as part of the study. Some participants reached out to the researchers directly while some have come through Stanford's sports medicine division, among other sources. All participants receive a detailed workup as well as sequencing at Stanford as part of the study. "Even if they come in from other places, we want them to be seen at Stanford as part of this, at least for this first trial," Bustamante said. Also, "because genetic data [is] being returned to participants, we have genetic counselors involved," he added.
The ELITE study is one of several projects at Stanford that use SAP technology and are the result of an ongoing collaboration that began roughly three years ago.
"We basically came to them and said 'honestly, sequencing genomes is no longer the bottleneck, we are sitting on tons of sequenced genomes, we just can't process them fast enough and really extract meaningful information'," Bustamante told GenomeWeb. Furthermore, "if we could aggregate all this genomic data with EHRs and wearables and whatever else is coming down the pike ... what are the technical hurdles that we'd need to overcome and would it give us a real-time system that we could query and ask questions that we didn’t know we had?"
As part of the partnership, SAP has provided funding for some Stanford projects. For example, funds from the company were used to sequence the genomes of some of the athletes enrolled in the ELITE study. Company researchers have also worked with researchers at the university to build software that supports these projects. "It's been a very iterative process," Bustamante said. "We started with the [SAP] innovation center, sketched out what we want it to look like, they came up with a prototype, [and] we suggested some changes."
The list of projects that use SAP infrastructure include one with Stanford School of Medicine to use the Hana platform to study the effects of mutations on cardiovascular diseases, for example, as well as to research how environmental exposure, behavior, and other factors contribute to disease susceptibility.
A more recent system involving SAP Foundation for Health is being developed in collaboration with Stanford's Clinical Genomics Service which, as the name implies, aims to provide genome sequence analysis and interpretation to physicians to help provide more personalized care for patients — this service also handles the sequencing for the ELITE study.
SAP's Singh explained that this will provide an automated system for filtering input lists of genetic variants down to the most likely harmful mutations based on effects such as loss of function as well as on ACMG guidelines. In addition to their top-ranked variants, patient pages within the system provide information on clinical timelines, relevant pharmacogenomics information, as well as ancestry information. Besides genomic and clinical data collected by healthcare providers, the developers also hope to be able to pull in information from wearables such as heart rate monitors and other devices.
One million cancer patients
At the SAP event, clinicians from the American Society of Clinical Oncology discussed the society's collaboration with SAP to develop a platform for combining clinical information in the context of cancer.
Earlier this year, SAP and ASCO announced their plans to develop a health information technology platform based on Hana called CancerLinQ. The platform is intended to combine de-identified data from millions of electronic health records, including molecular information collected as part of care or treatment, from different oncology practices and cancer centers. This information can be used to guide cancer therapy decisions and would also support efforts to develop personalized treatments.
During his presentation at the meeting, Clifford Hudis, a member of the CancerLinQ board of governors and past ASCO president, said that the organization hoped to have 1 million patient records in the system by June 2016. So far, the system contains records from roughly 500,000 patients collected by 350 physicians in 15 practices, he said. CancerLinQ CEO Kevin Fitzpatrick told GenomeWeb at the meeting that over 150 practices have expressed interest in contributing data to the system. "It's a testimony to the desire that physicians and healthcare teams have to get insight into what's really happening in their practices in novel ways," he said.
The data collected so far resides in a private cloud hosted by SAP for ASCO where it's currently being processed and cleaned before it will be made available. Ultimately, "the idea is that as we routinize data collection from various EMR systems and move that into production mode, it will be immediately available to an individual practice," Fitzpatrick said. ASCO and CancerLinQ don't provide specific guidelines for collecting data, instead allowing respective practices to use their own workflows and processes. "We do think that as we go further into this, greater standardization will occur as people strive to work with this data, understand this data, and as this data gets more heavily utilized when it comes to payment reform and other sorts of management systems," Fitzpatrick said.
The CancerLinQ project will roll out in three major phases, according to Fitzpatrick. For the first and current phase, the researchers are focusing on phenotypic descriptions of the diseases including details of organs and systems affected, patient symptoms, as well as therapies used. The second phase will focus on genotypic descriptions of the various cancers and decision support tools that will help providers make informed decisions about patient care. A third phase will focus on combining phenotype and genotype information with self-reported data from the patients themselves.
"The absolute bedrock intent of CancerLinQ is that it's all comers ... it represents everyone that walks through the door, there is no curation of that data, there is no cherry-picking of cases," Fitzpatrick said. "The idea is that we'll follow them longitudinally to understand what the short- and long-term consequences of care are."
That's particularly important because as cancer care improves and survivors live longer, some of the off-target effects of therapy are starting to show. For example, some patients who have had radiation therapy or some chemotherapy regiments have a propensity for heart disease and so there is increasing interest in cardio-oncology as a discipline, Fitzpatrick said. CancerLinQ makes it possible to follow these patients longer and look for signals that could point to the rise of these conditions and help physicians modify treatment regimens.
As part of efforts to prepare for later phases of CancerLinQ, ASCO has set up a number of committees to begin thinking about needs and potential issues that could crop up before SAP begins writing code. For example, there is a patient advisory committee whose goal is to ensure that there is a feedback loop to patients, meaning that they benefit from contributing their data to the system and are aware of ways to engage with the project, Fitzpatrick said. There's also a physician-led committee that's been set up to deal with things like data access issues.
"We want those deliberations to begin now so that when we begin to write specs for the developers, we can be confident that we understand what we are asking for and what the beneficial outcomes will be," Fitzpatrick said.