The aim of individualized treatment is to improve both quality of life and care while at the same time cutting costs. During times of tremendous burden of care, we are also weighing how the hospital’s resources can be put to the best use without cutting back on patient safety. The problem in the case of febrile neutropenia is that until now, there has been no useful tool with daily practice in mind that could enable us to classify patients with solid tumors based on risk of complications.
Two different approaches had been put forth: validated models or scores and exclusion criteria used in clinical trials. Both are recommended by clinical practice guidelines, such as the NCCN and ASCO guidelines, although the optimal method was still not clear at present. Moreover, the problem was that the two existing validated models, Talcott and MASCC, are generic classifications, developed in heterogeneous samples that included acute leukemias and bone marrow transplants. These two scenarios have a diagnostic and treatment approach that differs with respect to solid tumor. In fact, the FINITE study has shown that this mixture decreases the ability to discriminate within specific groups of patients, such as outpatients with solid tumors, the group of interest to the medical oncologist.
There were also other methodological concerns that hampered the applicability of the MASCC model in our daily practice. For example, the explanatory co-variant carrying the heaviest weight in the MASCC model is hypotension, which, on its own is also the most common output variable ("the prediction") in that same model. It is also an exclusion criterion for management as a low risk patient. In contrast, the “inpatient” variable in the same model, doesn’t really make a lot of sense for the physician who works in the Emergency Department, etc...
Because of its low specificity, MASCC model predicted only one out of every three serious complications in our series, which is fairly representative of the type of patients treated in routine medical practice in Medical Oncology departments. In fact, most randomized clinical trials that compared oral vs. intravenous antibiotic treatment, at home or in the hospital, did not use these validated scales as the single selection criterion. Oddly enough, they were based on pragmatic selection criteria.
Clinical stability has been objectively defined in these clinical trials and there is currently consensus in clinical practice guidelines as to which patients should initially be considered high risk, based on their history and a basic, clinical examination. In general, we are referring to patients with acute organ failure (renal, cardiac, respiratory) or decompensation of chronic insufficiency, septic shock or hypotension (systolic pressure <90 mmHg), severe infections (pneumonia, cellulitis >5cm, typhlitis, enteritis grade 3-4, appendicitis, meningitis, and pyelonephritis), and other serious complications constituting an admission criterion by themselves (pulmonary thromboembolism, arrhythmias and bleeding). These patients must always receive maximum support. However, for the physician who is making the decisions, it may be more important to be able to identify the patients without evident criteria of severity, but who are at potential risk for severe complications. The CISNE calculator has been developed specifically for this population.
Thus, the issue as to how to optimally select patients has not been established. On the one hand, there is the difficulty in “real clinical” practice in perceiving early, subtle signs of severity, due to the depressed inflammatory response in patients with cancer and neutropenia. On the other hand, most of the analyses of prognostic factors in patients considered to be clinically stable and who had participated in clinical trials were made with relatively small sample sizes in comparison with the recruitment capacity of the participating sites, which speaks probably to non-consecutive recruitments.
We believe that the irregular patient inclusion in clinical trials, physician or patient criterion, limits the interpretation of the selection criteria, often even stricter than what is openly stated. A clinical trial is therefore not the best scenario in which to validate an empirical risk criterion, since not all patients have the same likelihood of being included, based on their characteristics. In these conditions, a prospective, multicenter observational study, such as the FINITE study, based on consecutive cases in clinical practice, provides useful, relevant information.