Feline Pancreatitis and fPLI - Making Sense of Sensitivity and Specificity
Joanne Lewis BSc BVMS (Hons) MRCVS
As a vet experienced in the pressures of working in general practice, I am sure I am among friends when I confess that remembering the difference between sensitivity and specificity was not high on my list of priorities! I am not about to launch into statistics now either as most of you will probably be reading this in your two second lunch break. What I will say though, is that believing that the test I had requested from the lab was indeed going to be of diagnostic value (ie worth spending the owner's precious last pounds on), was and still is very important to me. Afterall, there is nothing worse than convincing a client to let you run a test and then being presented with an equivocal or unexpected result at the end of a busy consulting block on a Friday night!
There are certain assumptions that are made when a result arrives in your in-tray. A degree of trust comes into play - as a general practitioner you cannot be in control of how the test has been run. It is the clinical pathologist's role therefore to know the limitations of the procedures used and give you guidance on the interpretation of the results produced. Believing in a diagnostic test often comes down to facts and figures - sample size, sensitivity, specificity etc.
Getting Down to Specifics with fPLI
Whether it be chronic and mild, or acute and severe, feline pancreatitis cannot be diagnosed based solely on clinical, clinicopathologic or imaging findings. Affected cats are present with notoriously vague symptoms and often keep even the best of clinicians on their diagnostic toes! It is vital that we understand the applications and limitations of fPLI before convincing ourselves that it is the answer we have all been seeking for all our complex feline medicine cases.
How many "off colour" or "not quite right" cats come through the door every day in general practice? When extracting a history from owners, most will agree that their cat has been lethargic, inappetant to a degree and maybe having lost some weight. If you are lucky, the cat is kept indoors and the owner may have noticed the odd vomit or some inappropriately located diarrhoea. These sickly cats often don't help us clinically, often showing nothing more than a degree of dehydration. Acute and severe cases tend to show similar but more dramatic signs than the milder, often chronic cases. The severe cases are typically that bit "sicker" - a statement that I agree is almost as vague as the symptoms themselves.
Yes, these cats may well be showing symptoms of pancreatitis, or indeed a multitude of other complex feline medicine differentials! When you then consider that some of these conditions (eg lymphoma, IBD, cholangiohepatitis, hepatic lipidosis etc) can occur on their own or in conjunction with pancreatitis it suddenly becomes clear why feline pancreatitis is far from easy to diagnose. You may eventually reach a point in your diagnostic plan for these cases, where you would consider whether running a fPLI would be helpful - if not to diagnose pancreatitis, at least to exclude it.
On first glance the fPLI test would seem to have reasonable diagnostic value with figures quoted as "100% sensitivity in cats with moderate/severe pancreatitis" and "100% specificity in healthy cats". In other words, fPLI would currently seem quite reliable in diagnosing most acute cases of pancreatitis (as they tend to be more severe). Furthermore it would seem that these figures show that it would be very unlikely that fPLI would misdiagnose a normal, healthy cat with pancreatitis.
In reality however, normal/healthy cats don't usually present for diagnostic workup. We are much more likely to be presented with a cat that has those typical vague clinical signs mentioned previously and not unique to any particular condition. It makes more sense therefore, that the fPLI test's reliability would be better assessed on its performance in this more realistic and practical context.
Disappointingly, studies released so far are not overly convincing for fPLI having a reliable role in the detection of the most common form of feline pancreatitis - chronic, low-grade pancreatitis. The studies currently available suggest that in a sample of 13 cats with known chronic, low-grade pancreatitis on histopathology, the fPLI test only showed ~50/50 chance of providing a correct diagnosis. Based on these figures, in practice this means that when using fPLI in mild pancreatitis, for every cat you correctly diagnose, you miss another altogether.
In the same study cats were defined as "symptomatic" for pancreatitis if they showed lethargy, anorexia, vomiting, diarrhoea or weight loss and hence could indeed be suffering from an extensive list of other differentials. The specificity in these cats is 67%, so one in three cats is a false positive and hence is mistakenly identified as having pancreatitis by the fPLI test. These statistics are based on the available research to date on three cats so the true specificity in a larger population of cats could be much better or worse - it is clear that more studies are needed.
The main useful application for the fPLI test is in cases which are likely to have moderate to severe pancreatitis. Again, this is based on a study that found raised fPLI in all of the five cats with acute, moderate to severe pancreatitis, hence the 100% sensitivity, as there were no false negatives in these cats.
In an ideal world we would all have the luxury of interpreting a fPLI result with a full histopathology report in front of us, neatly classifying the type/severity of pancreatitis present so we would then be able to gauge the chances of our result being accurate; however in reality we rarely have this option and hence we are unsure whether the diagnosis is in fact pancreatitis, let alone what type it is.
The sensitivity and specificity of feline PLI increases with the severity of the pancreatic inflammation. Current research suggests that fPLI may not be helpful for cases with only mild pancreatic inflammation or those cats with non-pancreatic disease that may be showing similar signs as those with pancreatic disease. It is vital therefore to understand the limitations of fPLI before convincing ourselves that it is the answer we have all been seeking for all our complex feline medicine cases. It is all too easy to simplify the diagnosis of a still incompletely understood feline disease. It would appear that many further studies are required before fPLI can be heralded as the solution to those tricky "out of sorts" cats.