[ad_1]
The precise choice of treatment for breast cancer depends on the condition of the hormone receptors (for estrogen and progesterone). Their conventional determination by immunohistochemistry (IHC) is associated with a certain error rate, which can be reduced by adding genomic data. Even conventional statistics can provide a noticeable improvement, but now it is possible to use decision theory to optimally combine diagnostic results, especially when they are contradictory. This is the result of a recent study conducted by MedUni Vienna under the direction of Wolfgang Schreiner of the Center for Medical Statistics, Informatics and Intelligent Systems (CeMSIIS). The methodology has applications far beyond breast cancer and can be deployed in all circumstances where it is necessary to draw conclusions from many results at the same time, even if the results are contradictory.
Schreiner explains, “Driverless cars use multiple sensors to check if they can move freely. It is possible that one of the sensors detects an obstruction and calls for emergency braking, while another sensor does not detect any danger in the same situation. What are you doing then? There are two potentially bad decisions and each of them is risky in a different way: if the car does not brake, even if it is necessary, there could be a serious accident. However, if the car brakes unnecessarily, there is a risk of a rear collision by the following vehicle with potentially less severe damage. “We are faced with a similar situation in the choice of treatment for patients with breast cancer, which must be adapted to the state of the hormone receptors on which it depends.
There may be times when the IHC test produces a “positive” result, while a second measurement method, such as gene expression, produces a “negative” result for the same patient. Again, there are two potentially bad decisions. Heinz Kölbl, from the Division of General Gynecology and Gynecological Oncology at MedUni Vienna, explains: “If you only offer hormone therapy, because you mistakenly think the patient is ‘positive’, although in reality she is ‘negative’, she will miss her life – It would be the greatest evil. If, on the other hand, a “false negative” patient receives aggressive chemotherapy instead of more conservative hormone therapy, she will experience unnecessary side effects. ”
So how do experts think to react to conflicting measurement results? “This is precisely where decision theory comes into its own,” Schreiner points out. Unlike conventional statistics, decision theory does not consider just a single number, i.e. the probability of an event (e.g., “ positive receptor ”), but also the probability of other possibilities (“ possibly receptor positive ” or “ definitely not receptor positive ” = “receptor negative”). This global vision improves the quality of the result compared to “conventional” statistics, in particular when several sources of results are simultaneously relevant. In a big data reuse study, decision theory has now been applied for the first time to the diagnosis of hormone receptors: to do this, the CeMSIIS team worked with Heinz Kölbl, Christian Singer and Cacsire Castillo-Tong of the general division of MedUni Vienna. Gynecology and gynecological oncology. IHC receptor status and the overall gene expression profile were jointly determined and analyzed for 3,753 breast cancer patients.
Decision theory works and warns
In the original studies, treatments were selected based on the status of IHC receptors. Even conventional statistics have shown that genomic data seemed to contradict the IHC (gold standard) results in some cases. It was precisely at this point that decision theory was used in place of conventional statistics, as in many cases it can generate an accurate overall result from conflicting individual results. Schreiner explains: “Decision theory also has the advantage that it“ realizes ”when it is uncertain: it then generates a clear result of“ undecidable. ”This in itself is important information, a warning for so. say.” This occurred in the case of 153 patients, indicating that potentially suboptimal treatment decisions were made based on the IHC data alone. In fact, this (small) group of patients had significantly lower survival than the remaining, much larger group with confirmed IHC status and therefore “correctly” selected treatments.
The effectiveness of decision theory has therefore been demonstrated on a simple example, but extremely clinically relevant, for precision medicine: “Modern medicine increasingly uses many sources of information, especially in the fields of laboratory, critical care medicine and genomics.Data from different sources must be combined to reach conclusions, so far often in SOPs (standard operating procedures) with conventional Yes / No or even intuitively.This is where decision theory offers a vast array of potential improvements.
###
Warning: AAAS and EurekAlert! are not responsible for the accuracy of any press releases posted on EurekAlert! by contributing institutions or for the use of any information via the EurekAlert system.
Source link
