We have all encountered aberrant bits of tissue (floaters) on microscopic slides and within paraffin tissue blocks. Most of the time these are from obviously different tissue types and pose no diagnostic difficulty, though they often serve for us to remind the gross room and histopathology staff of the importance of scrupulous cleanliness. Sometimes, of course, the aberrant tissue is of the same type as the proper specimen on the slide and things become more difficult. If the floater is only on one slide, it's a little easier to assess that it represents a contaminant, than if it is present in the paraffin tissue block. In the latter case it becomes a matter of evaluating how much this presumed floater matches the rest of the tissue on the slide. For example if the presumed floater is a small fragment of high-grade colonic adenocarcinoma, the rest of the specimen is entirely normal colon, and there was no clinical suspicion of malignancy, then things seem relatively straightforward.
Unfortunately, sometimes it's not that easy. One approach that is often of value is to closely examine the other specimens accessioned, grossed, and processed on the same day. Does the presumed floater closely match any of these cases? Was the original tumor particularly friable and prone to shedding small tumor fragments? Does the fragment make sense in the context of the rest of the specimen and the clinical history? It is rare that some combination of these approaches does not result in confidence that a contaminant does (or does not) exist. In a few instances, though, either repeat biopsy or molecular testing can be performed to sort things out, the choice depending on the ease of the former v. the cost and time for the latter.
The above refers to instances of focal contamination of an otherwise appropriate specimen. A related and potentially more difficult problem is one of COMPLETE specimen mix up. These may be obvious, as when the tissue on the slide and in the paraffin block don't match the specimen designation, or the diagnosis is clinically untenable. As above, comparison with other similarly processed specimens will often reveal the flip side of the mixup. For example the endometrial curetting from a recently pregnant 24 year old shows high-grade serous carcinoma and the endometrial curetting from a 70-year-old woman with bleeding shows a decidual cast. All of us in practice for a reasonable length of time have encountered such cases. At my own institution we have taken considerable pains to educate clinic staff about this problem and, for example, the danger of pre-labelling specimen vials; something that seems like a good, timesaving, idea but is a setup for a mixup.
Recognized tissue errors of contamination or mixup have been noted to occur in about 6% of cases. A reasonable and somewhat unsettling question is, "How often is there a specimen mixup or contamination and we don't even know it?" Until now this question of occult errors has been seldom considered and never directly addressed. However, a recent article in the American Journal of Clinical Pathology (Am J Clin Pathol 2013;139:93-100) addresses just this point based on a large series of 13,000 prostate biopsies for which DNA data and short tandem repeat (STR) analysis were available, comparing the prostate biopsy with a buccal smear from each patient. Cases in which there was any clinical suspicion of a mixup were excluded in order to assess the truly occult rate of error. Rigorous protocols were followed to avoid mixups in the testing itself by using forensic type collection methods, chain of custody, and appropriate retesting.
The authors considered complete specimen mixup as a "type 1" error, and tissue contamination as a "type 2" error. The results indicated that type 1 errors occurred in 0.26% of cases and type 2 errors were three times as common, occurring in 0.67% of cases. Taken together almost 1 in 100 cases had an unrecognized "tissue origin" error. The study did not assess how many of the type 2 errors had the potential for significant clinical impact. All of the type 1 errors had at least this potential. The type 1 error rates could not be correlated with the type of pathology laboratory or clinical setting. The type 2 error rates showed a correlation with the pathology laboratory setting but, as expected, not with the clinical setting.
This study is a fascinating and somewhat unsettling example of applying current molecular technology to address a important patient safety issue.