An extensive molecular cytogenetic characterization in high-risk chronic lymphocytic leukemia identifies karyotype aberrations and TP53 disruption as predictors of outcome and chemorefractoriness

We investigated whether karyotype analysis and mutational screening by next generation sequencing could predict outcome in 101 newly diagnosed chronic lymphocytic leukemia patients with high-risk features, as defined by the presence of unmutated IGHV gene and/or 11q22/17p13 deletion by FISH and/or TP53 mutations. Cytogenetic analysis showed favorable findings (normal karyotype and isolated 13q14 deletion) in 30 patients, unfavorable (complex karyotype and/or 17p13/11q22 deletion) in 34 cases and intermediate (all other abnormalities) in 36 cases. A complex karyotype was present in 21 patients. Mutations were detected in 56 cases and were associated with unmutated IGHV status (p = 0.040) and complex karyotype (p = 0.047). TP53 disruption (i.e. TP53 mutations and/or 17p13 deletion by FISH) correlated with the presence of ≥ 2 mutations (p = 0.001) and a complex karyotype (p = 0.012). By multivariate analysis, an advanced Binet stage (p < 0.001) and an unfavorable karyotype (p = 0.001) predicted a shorter time to first treatment. TP53 disruption (p = 0.019) and the unfavorable karyotype (p = 0.028) predicted a worse overall survival. A shorter time to chemorefractoriness was associated with TP53 disruption (p = 0.001) and unfavorable karyotype (p = 0.025). Patients with both unfavorable karyotype and TP53 disruption presented a dismal outcome (median overall survival and time to chemorefractoriness of 28.7 and 15.0 months, respectively). In conclusion, karyotype analysis refines risk stratification in high-risk CLL patients and could identify a subset of patients with highly unfavorable outcome requiring alternative treatments.


INTRODUCTION
Chronic lymphocytic leukemia (CLL) is a heterogeneous disease, running an indolent course in some patients and a clinically aggressive course in others [1,2]. Therefore, in clinical practice, prediction of outcome and response to treatment is important in an era in which several chemoimmunotherapy combinations and effective mechanism-driven treatments are available [3][4][5].
Comprehensive prognostic indexes including clinical and biological parameters were published [23,24] and a recent systematic review [25] recommended IGHV and FISH analyses as useful tests for all newly diagnosed CLL patients as they may identify those 40-50% of cases with unfavorable prognosis based on the presence of an unmutated IGHV gene and/or 11q22/17p13 deletion [25].
Recently, karyotype aberrations were shown to represent a strong prognostic factor [26][27][28]. Moreover, the complex karyotype emerged as an independent predictor of inferior time to first treatment (TTFT) and shorter overall survival (OS) in patients investigated at diagnosis [29] or at disease progression [30] and in relapsed/refractory CLL treated with ibrutinib [31].
We therefore investigated whether an extended genetic characterization, including karyotype analysis using novel mitogens and mutational screening by NGS, could predict outcome in high-risk CLL patients. By correlating genetic data with clinical and biological parameters we showed that karyotype represents an independent prognostic factor for TTFT, OS and time to chemorefractoriness (TTCR).

Patients, mutational analysis and cytogenetic data
The clinical and biological characteristics of 101 high-risk CLL patients are presented in Table 1

Correlations between mutational status, cytogenetic findings and clinical and biological parameters
The presence of gene mutations did not correlate with age, sex, Binet stage and CD38 positivity, while a significant association was found with unmutated IGHV status (p = 0.040) and a complex karyotype (p = 0.047; Tables 2 and 3). A complex karyotype (Table 3) was also associated with unfavorable FISH (p < 0.001) and TP53 disruption (p = 0.012). A trend for a significant association was observed with the presence of ≥ 2 mutations (p = 0.061) and with TP53 and ATM mutations (p = 0.080 and 0.058, respectively). A positive correlation was found between unfavorable FISH and unfavorable karyotype (p < 0.001). 11q22 deletions by FISH were never found in association with TP53 mutations or with TP53 disruption. TP53 disruption was associated with the presence of ≥ 2 mutations (p = 0.001) and mutated IGHV status (p < 0.001), the latter association being accounted for by the selection criteria adopted in this analysis, which included patients with unmutated IGHV and TP53 lesions. An overview of the distribution of mutations and principal biological findings according to cytogenetic results is presented in Supplementary Figure 1.

Correlations between mutational status, cytogenetic findings and clinical outcome
The median follow-up for these 101 CLL patients was 41 months. In univariate analysis (Table 4), a worse TTFT was associated with advanced Binet stage (p < 0.001), a complex karyotype (p < 0.001) and an unfavorable karyotype (p = 0.001; Figure 1A). No correlations were observed between TTFT and gene mutations.
By multivariate analysis (Table 6) an advanced Binet stage (p < 0.001) and an unfavorable karyotype (p = 0.001) predicted a shorter TTFT while TP53 disruption (p = 0.019) and an unfavorable karyotype (p = 0.028) were associated with a worse OS. A shorter TTCR (Table 6) was predicted, in multivariate analysis, by TP53 disruption (p = 0.001) and unfavorable cytogenetics (p = 0.025). When in the model of multivariate analysis we introduced the complex karyotype instead of karyotypic abnormalities, the complex karyotype was significantly associated with a worse TTFT (p < 0.001) and an inferior OS (p = 0.024) while a trend for significance was observed for a shorter TTCR (p = 0.077; Supplementary Table 4).
By combining TP53 disruption status and cytogenetic results, 9 patients showed the coexistence of TP53 disruption and unfavorable karyotype: this subset of very high-risk patients presented a median TTFT of 5.5 months, a median OS of 28.7 months and a median TTCR of 15 months (Table 7, Figure 2).
In patients with favorable or intermediate cytogenetic findings (Supplementary Table 5

DISCUSSION
IGHV status and FISH results represent robust prognostic tests and their inclusion in the standard prognostic work up of newly diagnosed and/or previously untreated CLL was recently recommended [25]. Indeed unmutated IGHV status and 11q22 deletion were recognized as independent predictors of an unfavorable outcome [23,24] while 17p13 deletion was aggregated with TP53 mutations in a category that received the highest risk score in the international prognostic index (IPI) of CLL [24]. Thus, based on IGHV status, FISH results and TP53 mutational analysis it is possible to identify a category of CLL patients with poor prognosis, representing approximately 50% of all CLL cases [13].
Karyotype analysis was reinforced as an important variable in CLL risk assessment [29,31,32,33] though it is relatively time consuming and needs standardization of culture conditions [34]. Because few data are available on cytogenetic and molecular findings in high-risk CLL, we perform an extensive genetic characterization of 101 CLL with unmutated IGHV and/or 11q22 deletion and/or TP53 disruption. These patients were diagnosed and followed at a single center for a median of 41 months over the last 10 years. Notably, our center has a > 90% capture of incident CLL cases in our region allowing for meaningful analyses of TTFT, TTCR and OS in a real world scenario. In this study, cytogenetic analysis was performed following DSP30/IL2 stimulation, thus reducing the rate of cytogenetic failure [28,34]. A complex karyotype was present at diagnosis in 20.8% of patients, while a favorable karyotype was observed in 29.7% of the cases. Similar data were observed in the CLL 11 trial when karyotyping was performed before treatment and chromosomal aberrations were found in 68.8% of 154 patients, with 19.5% of the cases showing a complex karyotype [30]. These figures, as expected, are higher than those observed in unselected series of patients that included also low-risk CLL and cases with stable disease [29,33].
We performed mutational screening with the Ion Torrent PGM, a very sensitive NGS platform, allowing for multiplexing of samples and gene targets in one experimental setup thus resulting in higher speed of analysis and lower costs [35]. Parallel sequencing of exonic regions in 20 CLL-related genes showed gene mutations in 56/101 (55.4%) cases by using a 5% cut off. Mutations were detected with a frequency ranging from 5.0 to 96.7% of the reads, showing cases with major clones that represent early leukemogenetic events and cases with minor clones that represent late-appearing aberrations [36,37]. In this series, the frequency of mutations involving TP53, NOTCH1, SF3B1, ATM and BIRC3 genes (15.8, 10.9%, 10.9%, 7.9% and 4.9%, respectively) reflected the high-risk genetic features of our patients. Similar incidences for these mutations were reported in patients enrolled in clinical trials at disease progression [13]. The frequency of mutations involving the other investigated genes was in line with published data using whole exome sequencing [38][39][40][41]. Interestingly, we observed that 25.7% of the cases presented more than one mutation. In the CLL11 trial, NGS analysis revealed mutations in 42 out of 85 analyzed genes, with 76.4% and 42.2% of the patients presenting one or ≥ 2 mutations, respectively [30].
Moreover, an unfavorable karyotype, defined by the presence of a complex karyotype and/or 11q22 and/or   17p13 deletion, was independently associated with a worse OS and TTFT, while TP53 disruption was associated with poorer OS. It is worth noting, however, that our patients may have received different lines of treatment including various chemotherapeutic agents, with or without rituximab and BTK inhibitors. These observations need therefore to be validated in larger series of patients and in clinical trials with homogeneous treatments. Interestingly, in the CLL11 trial a complex karyotype emerged as an independent negative prognostic factor for OS after frontline therapy [30] while in relapsed/refractory CLL treated with ibrutinib-based regimens, a complex karyotype was stronger than 17p13 deletion in predicting an inferior outcome [31]. A complex karyotype was also found to be associated with unmutated IGHV genes and aberrations of chromosome 17p and was identified as an independent prognostic factor for shorter TTFT in a series of 1001 cases [33]. Likewise, an unfavorable karyotype predicted a shorter OS and represented the strongest prognostic parameter for disease progression in unselected patients studied at diagnosis [28,29]. The mechanisms responsible for the unfavorable prognostic significance of the complex karyotype remain to be investigated, though in CLL it was shown that telomere shortening was associated with genetic complexity [42].
We also assessed the impact of karyotyping and gene mutations on TTCR. While confirming the negative role of TP53 disruption [36,37], we showed, for the first time, that patients with unfavorable cytogenetics had a shorter TTCR. The identification of patients that are likely to develop chemoresistance is very important, given the possibility to use alternative effective BCR-target treatments [43,44].
Nor the presence of gene mutations, nor single gene mutations correlated with TTFT, OS and TTCR, with the notable exception of TP53 mutations. This is not surprising if we consider that the high-risk genetic profile of our patients may mirror a genetic instability or dysregulation [45] that may overcome the prognostic impact of single gene mutations [12,13].
Interestingly our data showed that high-risk CLL patients can be further sub-classified into four different groups, based on TP53 disruption status and karyotype. Those patients with both unfavorable cytogenetics and TP53 disruption presented a dismal clinical course (median TTFT, OS and TTCR of 5.5, 28.7 and 15.0 months, respectively). This observation may justify a specific treatment strategy for this subset of patients [46]. By contrast, there are cases with TP53 disruption and favorable or intermediate cytogenetics that may display a less aggressive course of the disease [47]. Recently, it has been shown that also the percentage of positive cells by FISH may help to refine the prognosis of high-risk CLL patients [48]. By contrast, mutational analysis may refine the prognosis of high-risk patients without unfavorable or complex karyotype or TP53 disruption as shown by the negative prognostic impact of NOTCH1 mutations and of the coexistence of two or more mutations [49].
In conclusion, we described the results of a comprehensive analysis of chromosomal aberrations and gene mutations in high-risk CLL, providing the first demonstration that the cytogenetic profile was independently associated with a shorter TTFT, OS and TTCR. The introduction of this technique in future CLL trials seems warranted to identify those high-risk patients that could be considered as ideal candidates for consolidation treatments or novel treatment combinations [50][51][52][53]. Larger series of homogenously treated patients and with longer follow-up could confirm this observation.

Patients
One hundred-one untreated CLL patients with unmutated IGHV gene and/or 17p13/11q22 deletion by FISH and/or mutated TP53 gene (here referred to as high- risk CLL) were included in this analysis. TP53 disruption was defined by the presence of 17p13 deletion by FISH and/or TP53 mutation by NGS. These patients belong to a consecutive series of 200 patients diagnosed and followed between 2007 and 2014 [49]. All patients were diagnosed according to NCI criteria [54]. Only patients with a Matutes immunophenotypic score [55] ≥ 3 (i.e., typical CLL) were included. CD38 and ZAP70 were tested on peripheral blood (PB) cells, as described [56]. When needed, mantle cell lymphoma was excluded by cyclin D1 evaluation. The study was approved by the local ethics committee. Indications for treatment included: increased white blood cell count with < 6 month lymphocyte doubling time, anemia or thrombocytopenia due to bone marrow infiltration or autoimmune phenomena not responding to steroids, and clinically significant disease progression in the Binet staging system. Fludarabine or bendamustine containing regimens, with or without rituximab were used as first-line treatment in fit patients; chlorambucil with or without rituximab was used in elderly and/or unfit patients according to the treatment policy adopted at our center. Since 2015, ibrutinib or idelalisib plus rituximab were offered to relapsed/refractory patients.

Cytogenetic and FISH analyses
Interphase FISH was performed on PB samples obtained at diagnosis using probes for the following regions: 13q14, 12q13, 11q22/ATM, 17p13/TP53 (Vysis/ Abbott Co, Downers Grove, IL) as described [57]. Each patient was categorized into a FISH risk group according to the following classification: favorable group Cytogenetic analysis was performed on the same samples used for FISH analysis using CpGoligonucleotide DSP30 (2 μmol/l TibMolBiol Berlin, Germany) plus IL2 (100 U/ml Stem Cell Technologies Inc., Milan, Italy) as described [57]. Whenever possible, at least 20 metaphases were karyotyped and each patient was categorized into a cytogenetic risk group according to the following classification: favorable group (isolated 13q14 deletion or normal karyotype), unfavorable group (deletions of 11q22 or 17p13, or complex karyotype, ie, at least three chromosome aberrations); intermediate group (all other karyotypic abnormalities).

IGHV analysis
IGHV genes were amplified from genomic DNA and sequenced according to standard methods with the cut-off of 98% homology to the germline sequence to discriminate between mutated (< 98%) and unmutated (≥ 98%) cases, as reported [58].

Next generation sequencing
NGS analysis was performed on the same samples used for FISH and cytogenetic analyses. In all samples, the percentage of CLL cells was over 90% as assessed by flow cytometry analysis. Agilent HaloPlex Target Enrichment kit (Agilent Technologies, Santa Clara, CA, USA) was used to  . To identify pathogenic variations, mutations that did not affect the protein coding regions (intronic, 3ʹ and 5ʹ UTR variations, silent exonic mutations and polymorphisms) were filtered out; insertions and deletions belonging to homopolymeric regions were removed, because sequencing error rate is high in these regions [49].

Statistical analysis
The Fischer's exact test was applied for categorical variables. TTFT was calculated as the interval between diagnosis and the start of first line treatment. OS was calculated from the date of diagnosis until death due to any cause or until the last patient follow-up. Refractory disease was defined as treatment failure (stable disease, nonresponse, progressive disease, or death from any cause) or disease progression within 6 months from antileukemic therapy [54]. TTCR was measured from date of first-line treatment until date of refractoriness to fludarabine/bendamustine-based regimens, date of alkylator refractoriness in patients who had never been exposed to fludarabine/bendamustine, death, or last follow-up [59]. Survival curves were compared by the log-rank test. Proportional hazards regression analysis was used to identify the significant independent prognostic variables on TTFT. The stability of the Cox model was internally validated using bootstrapping procedures [11]. Statistical analysis was performed using Stata 14.0 (Stata Corp, College Station, TX).