Molecular testing is advancing more rapidly than laboratory information systems (LISs) are able to keep up.1 Molecular laboratories and methods are unique in many ways and would benefit from a number of specialized tools to increase efficiency, ease workflow and improve overall patient care. As described in part 1, molecular laboratories that re-purpose their current general LIS to meet their needs often run into a number of stumbling blocks when it comes to reporting and data retrieval. These stumbling blocks could be removed through judicious information system design.
Described here is how information systems could be used for decision support and handling of high-throughput data in clinical laboratories as well as how appropriate storage, additional functionality and interface capability could be used in the future to aid minimal-risk research.
Clinical decision support (CDS) is a relatively new area in medical informatics and means the use of information systems to aid the healthcare provider in medical decision making. The vast majority of current literature surrounding CDS relates to prevention of medication errors, but a growing number of papers on CDS are focusing on laboratory testing. CDS components to LISs provide assistance to:
the pathologist or scientist when interpreting the test results.
Because molecular methods are relatively new, providers more often need CDS to know which test to order (e.g., determining whether to test for the BCR/ABL1 translocation by FISH or reverse transcriptase PCR). This type of CDS is now incorporated into many electronic medical records, but CDS and associated alerts for the laboratory staff, scientists and pathologists are still in a fledgling state.
As an example, laboratory staff that register a new paraffin block specimen into the LIS are prompted to enter the type of fixation; the system sets off an alert if the fixation took place in B5 or Bouin's fixative, both of which are known to prevent molecular analysis because of their destructive qualities on nucleic acids. Another example is an alert triggered when the recorded amount of nucleic acid in the sample is not consistent with the performance of the amplification controls. Specimens with adequate nucleic acid but little to no amplification of control genes could signify the presence of an inhibitor in the sample.
Alerts for failed controls are similarly used by the interpreter. If the decision is made to repeat the test on the same sample, then laboratories without specialized information systems may send a technologist to visually examine the specimen or check a log of nucleic acid quantities to determine if enough of the sample remains to perform another analysis.
Newer LISs specific to molecular laboratories include tracking of aliquots of extracted nucleic acid so this information is more easily retrieved. Keeping up with the storage locations of specimens in a molecular laboratory also has been challenging. Some information systems are now including storage tracking modules for all laboratories, enabling technologists to make the best use of their storage space by allowing them to store the samples in any location. For example, if samples are no longer stored next to samples with similar case numbers, technologists will be less likely to choose the wrong sample for testing.
In the future, genotyping of single nucleotide polymorphisms (SNPs) may be routine for specimen tracking and identification in molecular and other laboratories. Accordingly, it will be important for information systems to be able to handle and manage this data easily.
Easy retrieval of data is key for both clinical patient care and research. Preserving molecular data in an easily searchable format dramatically improves one's ability to mine the data for significant associations, and the importance of having retrievable data in the discovery process has been demonstrated by the contribution of bioinformatics to the human genome project.2 The combination of already existing molecular data, patient history and outcomes in a comprehensive electronic medical record can provide a wealth of information data from which new disease associations, diagnostic methods and therapeutic interventions can be derived with minimal to no risk to the patient.3
Improving data access and retrieval may have the greatest impact in genetic disorders, both inherited and acquired. Many inherited genetic disorders are relatively rare, so identification of a significant number of patients and their specific disease-causing mutations is difficult, which makes drawing any conclusions about their prognostic, diagnostic or therapeutic significance almost impossible. Molecular LISs could resolve some of these issues by communicating diagnostic and historical information to an already established genetic registry for the specified disease via a standardized interface (provided that the patient has given his/her consent). For families already suffering from the disease, this information could enable the associated researchers to determine if a family has a shared deleterious mutation in the gene tested (such as a BRCA1 or BRCA2 mutation for breast cancer), a shared deleterious mutation in another gene or shared environmental factors that predispose to disease.4
Sequence variations of unknown clinical significance are detected all the time, and for patients in whom a sequence variation of unknown clinical significance has been found, these registries could help identify other patients with identical or similar sequence changes and examine the patients' collective phenotype (also known as phenomics) to help determine whether the sequence variation is benign or disease-causing. As described in part 1 of this article, standards for storage and communication of data are needed for basic laboratory operations, but these standards could also include specifications for communication of the LIS to a genetic registry.
A few vendors have designed LISs to meet the needs of molecular laboratories, but even these systems will have to adjust to new technologies either currently in use or just on the horizon for use in clinical patient care. Magnetic bead technology, nucleic acid microarrays, nanotechnology and proteomics are just a few examples of high-throughput molecular analyses.5,6 Given the large amount of data generated by each one of these assays, the analysis, interpretation and selection of data for transfer from the instrument and into a molecular LIS becomes even more critical as the data starts to be used for clinical patient care. Standards for molecular information should not forget to include these types of assays even though many are not yet ready for prime time.
Molecular laboratories have a number of unique information needs that can be met with the use of a well-designed molecular information system. Individuals and institutions purchasing such systems must keep in mind their laboratory's current testing menu and workflow as well as anticipate future technologies to be incorporated into the laboratory.
Dr. Carter is director of Pathology Informatics for Emory University Hospital in Atlanta, GA.
1. Smith TJ. Molecular diagnostics: Outpacing informatics solutions? ADVANCE for Administrators of the Laboratory June 2007;16(6):58.
2. Elkin PL. Primer on medical genomics part V: Bioinformatics. Mayo Clin Proc Jan 2003;78(1):57-64.
3. GenBank. National Institutes of Health [website]. September 26, 2006. Available at: http://www.ncbi.nlm.nih.gov/Genbank/index.html. Accessed June 27, 2007, 2007.
4. John EM, Hopper JL, Beck JC, et al. The Breast Cancer Family Registry: An infrastructure for cooperative multinational, interdisciplinary and translational studies of the genetic epidemiology of breast cancer. Breast Cancer Res 2004;6(4):R375-389.
5. Hanai T, Hamada H, Okamoto M. Application of bioinformatics for DNA microarray data to bioscience, bioengineering and medical fields. J Biosci Bioeng May 2006;101(5):377-384.
6. Yu J, Yu J, Almal AA, et al. Feature selection and molecular classification of cancer using genetic programming. Neoplasia April 2007;9(4):292-303.