Droplet digital polymerase chain reaction for the assessment of disease burden in hairy cell leukemia

Abstract BRAF V600E mutation is the pathogenic driver of hairy cell leukemia (HCL) found in the vast majority of cases both at onset and during recurrences. The identification of the mutated allele in blood and marrow correlates with the presence of neoplastic cells and can be considered a marker of active disease. Likewise, the absence of the mutation after treatment may indicate a state of deep response. The BRAF V600E burden was measured by droplet digital polymerase chain reaction (ddPCR) and expressed as fractional abundance in 35 HCL patients at different stages of disease (onset, relapse, complete response [CR] after treatment, long‐term remission) in peripheral blood and/or bone marrow (when available). Mean values of fractional abundance for patients at diagnosis, relapse and response, respectively, were 12.26%, 16.52% and 0.02% in peripheral blood and 23.51%, 13.96% and 0.26% in bone marrow. Four patients out of 6 evaluated at response were molecularly negative for BRAF V600E in peripheral blood. Mean fractional abundance in peripheral blood tested in 14 patients with long lasting CR was 0.05%, and 10 patients were BRAF V600E negative. These preliminary results suggest that ddPCR permits to assess the active tumor burden in HCL at different disease phases and support the hypothesis that some patients in CR qualify for a molecular CR.

although with the invariable persistence of BRAF V600E . 3,4 This mutation is absent in other lymphoproliferative diseases that may mimic HCL, such as splenic lymphoma with villous lymphocytes and splenomegaly or splenic red pulp lymphoma, thus it is characteristically a hallmark of this disease. 1,5 Finally, the aberrant BRAF protein encoded by the mutated allele, has proven to be a suitable target for treatment by means of orally available inhibitors (vemurafenib above all), to be applied alone or in combination with other targeted agents like rituximab. [6][7][8] BRAF blockade is capable of a rapid recovery of peripheral blood cytopenias as well as of the reversal of the hairy cell phenotype of the leukemic cell. 6 The identification of the mutated allele in peripheral blood (PB) and bone marrow (BM) by means of polymerase chain reaction (PCR) or immunohistochemistry on decalcified tissue obtained trephine biopsy is of importance for diagnosis, although not indispensable. [9][10][11][12] Likewise, a positive immunohistochemical reaction on marrow tissue is suggestive of disease persistence after an active treatment, 13 although currently applied response criteria only rely on the resolution of peripheral blood cytopenias and of marrow lymphoid infiltrate in hematoxylin-eosin staining to assess the depth of the achieved response. 14,15 Internationally acknowledged guidelines do not currently recommend an assessment of minimal residual disease (MRD) in patients otherwise in complete response (CR), as its impact on patients' outcomes and medium-to-long-term prognosis has never been determined. 12 Nevertheless, MRD study is now an active field of research in HCL, and the application of immunohistochemistry thresholds for PAX5, CD20, DBA.44, VE1 or ANXA1, as well as flow cytometry assays including the four HCL markers CD11c, CD25, CD103 and CD123 antibodies have been proposed, but never formally validated. 12 The molecular assessment of the BRAF V600E burden can be an alternative option to detect MRD in residual disease, as well to confirm diagnosis at onset or relapse. Droplet digital PCR (ddPRC) has proven its accuracy in several hematologic disease, including chronic myeloid leukemia and HCL as well, and may be proposed as a tool to detect the presence of the mutation in an investigational setting. 16 We have implemented a molecular assay based on the use of ddPCR on PB and/or BM in patients with HCL at different phases of their disease, which may individuate a minimal disease burden in patients achieving a CR and that allows monitoring of all CR patients in the post-treatment phase. Here we present the preliminary results of this exploratory analysis on patients recently treated at our Institution because of a new diagnosis of HCL or relapse, as well as followed-up after having obtained a CR several years before. 17,18 2 | METHODS

| Study objective
Given the importance of the BRAF V600E mutation in the pathogenesis of the disease, being present in the vast majority of HCL cases both at onset and at relapse, we have hypothesized that the detection of the mutated allele in either PB or BM could correlate with the presence of neoplastic cells, indicating an active disease, regardless the severity of symptoms or the depth or PB cytopenias. On the other hand, we also postulated that the absence of the mutated allele after treatment (or even in patients with continuous CR) could indicate a state of profound response.

| Study overall conduct
In order to demonstrate our hypotheses, we have evaluated the

| Biologic material and ddPCR procedure
Study material consisted of 20 mL of PB collected in EDTA tubes and up to 6 mL of BM, when aspiration was possible. All patients underwent BM trephine biopsy both before and after treatment, which was required to assess the depth of response to treatment by means  Seven patients have been evaluated at response: importantly, we have selected only patients achieving a CR, as defined by the Consensus Resolution criteria (see Supporting Information). 14 All the patients evaluated at response have been also tested immediately before treatment, so that paired analyses could be available. Table 1 summarizes the typology of patients tested and the site of sampling. In addition, we have also tested 14 patients who have previously achieved a complete remission after only one line of treatment with cladribine, provided their duration of response was longer than 5 years. In total, 36 patients have been tested: unsurprisingly, 97% of them were males. The overall number of measurements was 56, given that many patients were evaluated more than once (with assays on both PB and BM, and at disease onset and response). Point values for each of the performed assay are represented in Figure 2 (with vertical axis logarithmic scale) and in Figure S1 (linear axis

| Patients with no response to therapy remain BRAF V600E -positive
Besides assays performed on patients in CR after treatment, three patients lacking response have also been tested as a positive control.
All of them had a BRAF V600E proven positivity at the time of diagnosis or relapse (thus are included in the data presented above). One of them was tested on both PB and BM, displaying a FA of 9.

| DISCUSSION
The Consensus Resolution criteria, along with recently updated guidelines for the diagnosis and management of HCL patients, rely on the resolution of all peripheral cytopenias and organomegaly, as well as on the disappearance of marrow lymphoid infiltration with hematoxylin-eosin staining, to establish the status of CR. 14,15 Immunohistochemistry may be a strategy to assess the persistence of a minimal marrow infiltrate: however, neither thresholds for positivity, nor the most appropriate immunohistochemical markers have been validated or widely acknowledged. 13,17,20,21 Likewise, flow cytometry appears a useful tool, but no consensus is established on which antibody panel needs to be used. 12,22,23 The application of molecular techniques is still matter of research.
The integration of the standardly defined CR category with MRD data assessment-regardless the method used-may help stratify patients in CR better: in other words, it can be hypothesized that patients in CR with persistent MRD may have a different long-term prognosis than CR patients with negative MRD findings. 23,24 Importantly, it is well known that some patients achieve and maintain a long-term response after only one treatment line, although the factors predicting such a good prognosis are still unknown. 18 It is much likely that long-term responders have achieved a much more profound response than what obtained by those suffering of early relapse, albeit categorized in any case as complete responders. In other words, the achievement of an MRD-negative status may translate into better prognostic rates and much longer treatment-free intervals.
Our preliminary results suggest that ddPCR permits to assess the active tumor burden in HCL at different stages of disease and support the hypothesis that some patients in CR qualify for a complete molecular response. The BRAF V600E quantified as FA is a good descriptor of disease burden at each phase of HCL natural history and discriminates between patients with active disease (i.e. those at onset and relapse, as well as those with less than CR after therapy), patients in CR with MRD and patients in CR free of the disease.
Beyond its utility in disease diagnosis, as proved by previous literature reports, [9][10][11] such an assay can be applied for: (i) response categorization, as it may integrate the traditional CR status with the addition of the molecular status of the patient, either BRAF V600E positive or negative; (ii) relapse assessment, in case of peripheral cytopenias not unequivocally attributable to an overt HCL recurrence; (iii) non-invasive follow-up in patients after active treatment, with possible influence on the prediction of a symptomatic disease relapse; (iv) evaluation of long-term responding patients, who may be considered cured in case of a negative assay, provided their follow-up is sufficiently long.
We found the FA value of 1% represents the threshold to discriminate between active disease (with need of treatment) and CR.
In other terms, basing on our data it emerges that all patients with active disease display an FA > 1%; conversely, patients in CR display an FA < 1%, with some of them being molecularly negative. None of the patients with active disease was BRAF V600E negative in our case series.
Although preliminary, these results offer a wider interpretation of the response status obtained after active treatment. The values of FA obtained can be integrated coherently in patients' clinical context, along with their PB counts and BM findings. Besides that, this study has some limitations. First, data are not conclusive on the comparability of PB and BM. This is mainly due to the small number of observations and to the fact that BM is not always easily aspirated in HCL because of reticulin fibrosis. Our data show a persistent minimal positivity of marrow BRAF V600E burden in CR patients, although markedly below the 1% FA threshold, suggestive of an incomplete clearance of BM infiltration. Secondly, we are unable to provide a clear interpretation of FA results falling in close proximity of 1%.
These may refer to patients much likely affected by active-although possibly not symptomatic-disease and need to be contextualized within the global clinical history and integrated with more data (e.g. closely repeated PB counts). Third, we lack the initial BRAF mutational status of all long-term responding patients, although they are supposed to be positive in the vast majority due to the extremely high incidence of the mutation in the HCL population at diagnosis. 1,3 Lastly, paired data (i.e. BRAF mutational status before and after treatment) are available in seven patients only, corresponding to those, who received treatment during the study time window.
These preliminary results suggest that ddPCR permits to assess the active tumor burden in at different disease phases and support the hypothesis that some patients in CR can reach a complete molecular response. Further studies are required in order to follow patients prospectively by sampling serially PB and BM at each stage of the disease and to provide validation of this technique.

ACKNOWLEDGMENTS
Open Access Funding provided by Universita di Bologna within the CRUI-CARE Agreement.