We report here important observations about diagnosis, treatment and prognosis of meningeal carcinomatosis in patients with advanced breast cancer. Our cohort appears representative of this clinical setting, with an important proportion of patients with very advanced breast cancer, as well as lobular carcinoma or triple-negative cancer subtypes, as already reported [1, 8, 10].
Meningeal carcinomatosis may frequently be clinically misleading. We describe here how a multidimensional approach may be necessary to confirm the diagnosis of meningeal carcinomatosis. A combination of MRI and biological analyses of the cerebrospinal fluid has been suggested as being the best approach [6], and almost all our patients underwent both procedures. Our observations are similar to the literature, with elevated protein levels in the CSF (> 50 mg/dl) in 56%-91% of patients (5). Diagnostic relevance of Cyfra 21 − 1 is confirmed, as elevated CSF levels of Cyfra 21 − 1 are associated with all other diagnosis features. This result is similar to several studies on MC from other primitive solid tumor, such as lung cancer [11], or head and neck cancer [12]. Abnormal baseline Cyfra 21 − 1 did not reach statistically significant prognostic value, although evidence suggests that after the start of IT treatment, elevated CSF Cyfra 21 − 1 is an independent adverse prognosis factor [3]. Statistical significance was reached for a baseline CSF Cyfra 21 − 1 value greater than 79 ng/mL, corresponding to patients in the upper quartile, which we used in our score. Strikingly, we overall show that less of 25% of patients had comprehensive features of meningeal carcinomatosis (Fig. 1a). Taken together, these results suggest that the formal diagnosis of meningeal carcinomatosis should rely on a combinatorial but not exclusive approach, harnessing clinical, biological and radiological methods.
We used widely described therapies for patients with advanced breast cancer, including intrathecal therapy [1, 3, 6, 9]. We focused the present analysis on early response to IT therapy, then on overall survival prognosis, in order to identify which patients could potentially benefit from extended intrathecal therapy. After one month of IT treatment, almost all patients experienced either an improvement in clinical symptoms which triggered early clinical evaluation as a potential prognostic factor, or a decrease in the presence of tumor cells in the CSF or in the Cyfra 21 − 1 CSF level, with any relevant correlation, in line with the current limitations in evaluating response in leptomeningeal disease [7]. Only 15 patients had a comprehensive response, none evaluable with magnetic resonance imaging. It is noteworthy that so far no standardized MRI evaluation of leptomeningeal disease has been developed, due to a very low interobserver agreement[7].
Based on these results, we constructed the prognostic univariate and multivariate analyses on individual response criteria. Median overall survival was 32.4 weeks (range 2.6–230), consistent with recent large scale studies[8]. Most interestingly, clinical evaluation at one month was the most statistically significant factor associated with overall and 24-weeks survival. Clinical response at one month retained its powerful prognostic value in all multivariate models. Specifically treating patients with intrathecal chemotherapy has been demonstrated to improve progression free survival and reduce the number of MC-related death events [4]. Other biologically sound parameters also attained borderline significance, such as triple negative subtype (which therefore was not retrieved from the analyses), CSF Cyfra 21 − 1 levels or presence of tumor cells, and were retained in the multivariate models. None of the features associated with the cancer history or burden of disease showed significant prognostic value, as opposed to previous studies which did not consider neither the early clinical evolution nor key MC specific biological parameters [1, 9, 13].
Several scores have been already proposed, including in very recent studies, to refine the prognostic evaluation of meningeal carcinomatosis, however with very limited discriminating properties [8]. We introduce here two important novelties to build four original prognostic scores. First, we based the scores on a combination of clinical evaluation and biological response criteria specific to meningeal carcinomatosis. Second, we considered both overall survival and the landmark 24-weeks survival, because of the poor prognosis of BC MC and for clinical relevance purposes. Regarding overall survival, the 3 classes score based on 1-month clinical response and baseline Cyfra 21 − 1 (in the upper quartile) identified a very poor prognosis subset of patients. Regrouping the poor and intermediate prognosis groups confirmed a clear identification of two different prognostic subgroups. The 24-weeks OS prognostic score also described very well a very poor prognosis population. This result appears clinically relevant and applicable, as it is based on two variables easily measurable in daily practice: clinical response and 1-month CSF tumor cells, whereas Cyfra 21 − 1 is not always part of the usual biological analyses.
The present report has several limitations. It is an observational, retrospective and monocentric work. Due to our very specific approach combining evaluation of the neurological clinical response by the same trained clinical team and original biological CSF analyses, it has not been possible to build a validation cohort. Of note, our population consisted exclusively of patients treated with intrathecal chemotherapy. This can be a recruitment bias, by selecting only patients fit enough to be treated at the onset of meningeal carcinomatosis, thus not representing the entirety of patients with advanced breast cancer and meningeal carcinomatosis.
In summary, and to the best of our knowledge, this report is the first to propose comprehensive prognostic scores for both long- and short-term overall survival in patients with advanced breast cancer and meningeal carcinomatosis, considered treatable by intrathecal chemotherapy. As there is still no standard of care for such patients, we suggest that simple features such as clinical evaluation and presence of tumor cells in the cerebrospinal fluid after one month of treatment, and baseline CSF level of Cyfra 21 − 1 could be used in order to adequately decide which patient will potentially benefit from a prolongation of intrathecal therapy.