Treatment Outcomes and Secondary Cancer Incidence in Young Patients with Hairy Cell Leukaemia

Summary

Repeated therapy of hairy cell leukaemia (HCL) with treatments that have potential long-term toxicities has raised concerns regarding increased risk for younger patients. We compared clinical outcomes and disease complications in 63 patients with HCL aged ≤40 years at diagnosis with 268 patients >40 years treated at Memorial Sloan Kettering Cancer Center. The rate of complete remission following initial therapy was 87% and 83% (P=0.71) and estimated 10-year overall survival was 100% and 82% (p=0.25) in younger and older patients, respectively. Younger patients required therapy earlier and had a significantly shorter time between first and second therapy (median: 63 v 145 months) (P=0.008). Younger patients required significantly more lines of therapy during follow-up. The 10-year cumulative incidence of secondary malignancies in young and old patients was 0.205 and 0.287, respectively (P=0.22). The incidence of secondary cancers in patients aged >40 years at diagnosis increased with the number of treatments for HCL (P=0.018). These results highlight that young patients with HCL have shorter responses to treatment and require more lines of therapy to maintain disease control, while attaining similar long-term survival. This has implications in the design of future clinical trials given our findings that secondary malignancies increase with more chemotherapy exposure.

Introduction

Hairy cell leukaemia (HCL) is a rare low-grade lymphoproliferative disorder, which occurs 5 times more frequently in men. The disease is diagnosed at a median age of 49–51 years (Else, et al 2009), significantly earlier than other lymphoproliferative disorders, such as chronic lymphocytic leukaemia. Age at diagnosis of HCL is not thought to impact the clinical presentation, treatment outcomes or disease complications; however, this has not been studied in detail. We sought to assess the impact of age at the time of HCL diagnosis on treatment outcomes and disease-related complications.

Presenting features of HCL include pancytopenia and splenomegaly in the absence of lymphadenopathy. Bone marrow assessment reveals diffuse infiltration by lymphocytes, which express CD20, CD11c, CD103 and CD25 in the classical type of HCL. The neoplastic cells harbour the activating BRAF V600E mutation; only expression of IGHV4-34 and the variant form of HCL have been shown to predict for absence of this mutation (Waterfall, et al 2014). Mutations in CDKN1B occur in 16% of patients and are not thought to impact prognosis, although their association with age at diagnosis is not known (Dietrich, et al 2015). Treatment with purine analogues (cladribine and pentostatin) results in a median relapse-free survival of 10.5 years following first therapy with shorter periods of remission following subsequent treatments (Else, et al 2009). In recent years, the addition of rituximab (Ravandi, et al 2011) to cladribine as well as BRAF targeted therapy with vemurafenib has resulted in favourable responses in patients with multiply relapsed or refractory HCL (Tiacci, et al 2015). HCL may be associated with a small increase in the incidence of solid organ malignancies and skin cancers, which may be related to chemotherapy exposure or directly due to the immune suppressive effects of HCL itself (Kurzrock, et al 1997). The effect of patient age has not been assessed as a factor for predicting secondary cancers in this disease. A study from our group showed that HCL treated after the introduction of cladribine did not measurably increase the incidence of melanoma beyond the expected population rate (Watts, et al 2015); however, patients were not stratified by age at diagnosis in this report, which may be relevant given the differences in lifestyle and cumulative ultraviolet exposure in younger individuals (Watts, et al 2015). The issue of skin malignancy was of particular concern in the phase 2 study of vemurafenib, which resulted in investigators limiting drug exposure to 6 months because of concerns that BRAF inhibition could, paradoxically, increase the growth of RAS-mutated cancers (Su, et al 2012).

Given the favourable treatment responses and survival of patients with HCL, the need for repeated therapy has raised concerns regarding the increased risk of secondary malignancy, and this may be of particular significance in young patients who are expected to live with the disease for a longer period of time. Young HCL patients were arbitrarily defined in a previous report as being below 40 years of age at the time of diagnosis (Rosenberg, et al 2014). We applied the same age cut-off to evaluate clinical outcomes and secondary cancer incidence of young patients amongst 331 consecutively treated HCL cases seen between 1983 and 2013 and followed in the institutional database at Memorial Sloan Kettering Cancer Center (MSKCC).

Methods

Sample population

We identified 331 patients in the institutional HCL database who were seen between June 1983 and December 2013. Diagnosis of HCL was confirmed by local review of diagnostic or relapsed bone marrow histopathology and immunophenotype. We applied the age cut-off of 40 years at disease diagnosis to define young patients in order to maintain consistency with prior published studies (Rosenberg, et al 2014). Sixty-three patients diagnosed with HCL at age ≤ 40 years were compared to 268 older patients (aged >40 years at diagnosis). Sensitivity analysis was performed to assess treatment responses and survival using disease diagnosis age cut-offs of 50 and 60 years. The study was approved by the MSKCC Institutional Review Board.

Clinical response criteria

Clinical complete response (cCR) was defined as the normalization of complete blood count (CBC) and resolution of splenomegaly and/or lymphadenopathy. Partial response (PR) was defined as a ≥ 50% improvement in the CBC values or splenomegaly compared to that at diagnosis, and stable disease (SD) was defined as a <50% improvement in CBC. Bone marrow assessment following therapy, including measurement of minimal residual disease (MRD), was infrequently performed, limiting our ability to incorporate bone marrow assessments in the response criteria definitions. Relapse was defined by the development of new or progressive cytopenia, recurrence of lymphadenopathy or splenomegaly or new clinical symptoms attributable to HCL. Cytopenia as an indication for initial therapy or re-treatment was defined as an absolute neutrophil count <1.0 × 10⁹/l, platelet count <100 × 10⁹/l and haemoglobin <100 g/l. Secondary malignancies were defined as any cancer diagnosed after the diagnosis of HCL and were classified into skin cancers (non-melanoma and melanoma) and non-skin cancers.

Time-to-event endpoints, including overall survival (OS), time to next therapy (TNT) and cumulative incidence rates (CIR) for secondary malignancies, were analysed on a subset of 173 patients with classical HCL who were seen at MSKCC within 3 months of disease diagnosis. Limiting inclusion of patients only to those seen at MSKCC within 3 months of HCL diagnosis was performed to reduce reporting and selection bias inherent in retrospective cohort analyses as well as to ascertain a representative sample of newly diagnosed cases. Patients with HCL variant were excluded due to established differences in clinical features and outcomes in this form of HCL (Matutes, et al 2001). Karyotype was assessed using G-banding karyotype analysis of bone marrow and BRAF mutation testing was performed on bone marrow using polymerase chain reaction (PCR) or immunohistochemistry. Tissue subjected to this analysis had >5% tumour involvement at the time of assessment.

We also performed a separate analysis of treatment responses stratified by age at diagnosis for patients treated with vemurafenib in the recent phase 2 trial conducted in the United States (Tiacci, et al 2015).

Statistical analysis

Fisher’s exact tests and Wilcoxon rank sum tests were used to compare the characteristics of patients aged ≤40 versus >40 years at diagnosis. Sensitivity analysis was performed to assess OS and time to next treatment using age cut-offs of 50 and 60 years at HCL diagnosis. The age cut-off of 30 years at HCL diagnosis was not assessed because there were too few patients diagnosed below the age of 30 years to allow meaningful statistical comparison. OS and TNT were estimated using Kaplan-Meier methods, and differences between groups were assessed using the log-rank test. OS was defined from date of HCL diagnosis until death, and patients who were still alive at the end of the study period were censored at date of last available follow-up. TNT was defined from date of start of first-line therapy until date of start of second-line therapy or death. Patients who did not receive a second line of therapy or died during the study period were censored at date of last available follow-up. The cumulative incidence of a second cancer and the cumulative incidence of skin cancer were estimated using competing risks methods, with death as the competing risk. Differences between groups were assessed using Gray’s test. All statistical tests were two-sided, and p<0.05 was considered statistically significant. Statistical analyses were performed in R (version 3.2.0; R Development Core Team, Vienna, Austria), including the “survival” and “cmprsk” packages.

Results

Sample population

We identified 331 patients in the HCL database who received treatment between June 1983 and December 2013 (Table I). The median age at diagnosis was 52 years, and the median follow-up was 69 months (range 0–423). Twenty-seven (8%) patients were treated prior to 1990 and received either interferon or splenectomy as first line therapy. Of these, 16 patients required re-treatment and 15/16 received cladribine, thus almost all patients received treatment with a purine analogue, which remains the current standard for initial therapy. Sixty-three patients (19%) were ≤ 40 years of age at diagnosis, with the youngest patient presenting at the age of 19 years.

Table I. Patient characteristics according to age at diagnosis.

331 Patients with HCL were reviewed at MSKCC between January 1983 and December 2013. The type of HCL could not be confirmed in 65 patients because full immunophenotyping and histology results were not available. BRAF V600E mutation status was only assessed in patients with classical HCL and only in patients with >5% residual disease as measured by flow cytometry or immunohistochemistry (in bone marrow). BRAF mutation was identified using polymerase chain reaction or immunohistochemistry.

	≤ 40 years of age at diagnosis	>40 years of age at diagnosis	P-value
N (%)	63 (19%)	268 (81)
Male, N (%)	48 (76%)	212 (79%)
Median age at diagnosis, years (range)	37 (19–40)	55 (41–84)
Diagnosis date range	1 January1974 – 20 January 2013	1 January 1973 – 15 December 2013
Median follow-up, years (range)	7 (0–35)	5 (0–33)
Disease type, N (%)
Classical HCL	50 (96%)	198 (93%)	0.54
Variant HCL	2 (4%)	16 (7%)
HCL type not determined	11	54
Time (months) from diagnosis to initial therapy (range)	0.4 (0–84)	1.3 (0–161)	<0.001
Median WBC count at diagnosis, × 109/l (range)	2.8 (0.5–100)	3.3 (0.7–35.9)	0.064
Abnormal karyotype	3/15 (20%)	7/90 (8%)	0.15
BRAF V600E	11/12 (92%)	34/36 (94%)	0.73

HCL, Hairy cell leukaemia; WBC, white blood cell

Pathology features

Diagnostic pathology features suggested a trend to more pronounced leucopenia in younger patients; the expression of HCL-defining CD markers was identical in the two groups. TRAP staining on bone marrow was positive in all patients aged ≤ 40 years with classical HCL and negative in 3 patients aged >40 years (Supplementary Table 1). There was no difference detected in the incidence of HCL variant in the two age groups (P=0.54). Karyotype was determined in 105 patients at diagnosis and was abnormal in 9%. Patients with HCL variant were more likely to have an abnormal karyotype (3/7) compared to those with classical disease (7/94) (p=0.020). At the time of classical HCL diagnosis, 2 patients age ≤40 years had abnormal cytogenetics, while 5 in the >40 years age group had abnormalities. Partial or complete deletion the long arm of chromosome 7 was the most common abnormality found at diagnosis and relapse, followed by deletions of chromosome Y, which was probably age-related as this change was only seen in patients diagnosed over the age of 40 years. At diagnosis, deletions of 7(q) were seen in 3 patients, all of whom were over 40 years of age at diagnosis. At the time of disease relapse, 3/10 evaluable patients aged ≤40 years had deletion of 7(q), and 3/20 patients >40 years of age had this abnormality. Assessment of BRAF V600E mutation was available in 12 young and 36 older patients with classical HCL (Table I). The mutation was assessed by PCR in 39 patients and by immunohistochemistry in 9 patients, all of whom had over 5% residual disease (measured by flow cytometry) in bone marrow or peripheral blood at the time of testing. There was no difference in the incidence of BRAF V600E mutation between the two age groups (P=0.73).

Therapy

81% of patients required treatment during the follow-up period with no differences in this rate between the two age groups (P=0.25) (Table II). 82% received cladribine first, splenectomy was used upfront in 17 (6%) patients and interferon was used first line in 14 patients. All but one patient treated with upfront splenectomy and one with interferon received this therapy prior to 1990. The overall response rate (ORR) was similar between the two age groups with 98% of young and 94% of older patients attaining a response to treatment (P=0.71). The cCR rate was also similar between the two groups. A single patient in the ≤40 years group who did not achieve a cCR following cladribine was found to have HCL variant, while the others all had classical disease. In the >40 years age group, 9 of the 30 patients who failed to achieve cCR with cladribine had HCL variant. The total number of individual courses of therapy required to maintain disease control was significantly higher in younger patients, with 49% needing 2 or more treatments compared to the 30% in the >40 years group (P=0.015). There were no differences in the types of salvage treatment used in the two age groups with purine analogues, splenectomy, vemurafenib, non-purine analogue cytotoxics, rituximab or treatment with other agents on other clinical trials used with similar frequencies.

Table II.

Treatment and response to initial therapy, stratified by patient age at diagnosis (N=331).

	≤40 years of age at diagnosis, N (%)	>40 years of age at diagnosis, N (%)	P-value
First Treatment			0.25
2CDA	40 (63%)	180 (67%)
Other	13 (21%)	34 (13%)
Never treated	10 (16%)	54 (20%)
Response to first therapy			0.71
cCR	39 (87%)	144 (83%)
PR	5 (11%)	19 (11%)
SD	0	7 (4%)
PD	1 (2%)	4 (2%)
Total number of therapies			0.015
0	10 (16%)	54 (20%)
1	22 (35%)	134 (50%)
≥2	31 (49%)	80 (30%)
Other therapies received during follow-up
Splenectomy	8 (13%)	21 (8%)	0.22
Vemurafenib	2 (3%)	10 (4%)	0.99
Ever treated on clinical trail	4 (6%)	13 (5%)	0.54
Deaths	4 (6%)	21 (8%)
HCL	1 (2%)	8 (3%)
Other cancer	3 (5%)	13 (5%)
Non cancer-related	0	0

2CDA, cladrabine; cCR, clinical complete response; PR, partial response; SD, stable disease; PD, progressive disease. HCL, hairy cell leukaemia.

Vemurafenib response

We compared outcomes of patients stratified by age treated on the US phase 2, vemurafenib study (Tiacci, et al 2015). Twenty-four patients were eligible for response evaluation; 5 patients were age ≤40 years at diagnosis and 19patients aged >40 years were assessed. The ORR was 100% in both age groups, while the complete response (CR) rate as defined by the study criteria was 80% in young and 32% in older patients (P=0.12). The estimated OS of young and older patients at 12 months was 80% and 100% and PFS at 12 months was 60% and 83%; the sample size precluded a statistical comparison of these endpoints. The median number of lines of prior therapy was 3 in each age group.

Overall survival and time to next therapy

Time-to-event clinical outcomes were determined on a subset of 173 patients with classical HCL who were seen at MSKCC within 3 months of disease diagnosis. Median follow-up among survivors was 42 months in those aged ≤40 years and 46 months in those >40 years of age (P=0.46). 130/173 (75%) patients required treatment for HCL during their follow-up. The median OS for the entire cohort was not reached, while estimated 5- and 10-year OS for patients >40 years was 93% and 82% (Figure 1) compared to 100% at both time points in younger patients (P=0.25) (Table III). Four patients with abnormal cytogenetics had similar 5-year OS to those with normal cytogenetics. 25 patients died during follow-up, 9 of these were due to progressive HCL and 8 of these were patients >40 years. 16 died due to other malignancies, with comparable rates in the two age groups (Table II).

Figure 1.

(1) Patients diagnosed with hairy cell leukaemia (HCL) at ≤40 years of age had a significantly shorter period between first and second therapy (p=0.008). (2) Overall survival from date of diagnosis stratified by age at HCL diagnosis showed a trend to improvement in patients aged ≤40 years at diagnosis (p=0.25). (3) Cumulative incidence of all second cancers stratified by age group showed a trend to more secondary cancers in patients >40 years of age at HCL diagnosis (p=0.22). (4) There was no significant difference in cumulative incidence rate of skin cancers by age (p=0.71).

Table III.

Overall survival assessed by age cohorts (N=173).

	N	5-year OS (95% CI)	10-year OS (95% CI)	P-value
Entire cohort	173	94% (89–98%)	84% (76–94%)
Age at diagnosis				0.25
≤ 40 years	28	100%	100 %
> 40 years	145	93% (87–98%)	82% (71–93%)
Cytogenetics
Normal	59	96% (90–100%)	83% (65–100%)
Abnormal	4	100%	NA

The TNT following initial therapy in patients aged ≤ 40 years was 63 months compared to 145 months in the older group (P=0.008) (Figure 1.1). Twenty-six patients required a second treatment with a median time to second therapy of 138 months (Table IV). By 5 years, 80% had not yet required a second therapy. The median TNT in 116 patients treated with cladribine was 138 months, while the median TNT in 8 patients who received pentostatin was 81 months (P=0.70). In the 26 patients who relapsed after second line treatment, the median TNT following the second therapy was not reached, and 80% had not yet required a third treatment after 5 years. The median TNT after first salvage with cladribine was 66 months, while the median TNT was not reached in the 4 patients who received combination cladribine and rituximab (Supplementary Table 2).

Table IV.

Time to second treatment stratified by age and initial therapy (N=130).

	N	Median TNT, months (95% CI)	5-year TNT (95% CI)	10-year TNT (95% CI)	P-value
Full cohort	130	138 (101, NA)	80% (72–89%)	57% (46–71%)
First treatment					0.70
2CDA	116	139 (101, NA)	80% (71–89%)	57% (45–71%)
Pentostatin	8	81 (NA)	86% (63–100%)	NA
Age at diagnosis					0.008
≤ 40 years	20	63 (35, NA)	52% (30–90%)	NA
> 40 years	110	145 (101, NA)	85% (77–93%)	60% (48–75%)

TNT, time to next therapy; 95% CI, 95% confidence interval; 2CDA, cladrabine

Sensitivity analysis using diagnosis age cut-offs of 50 and 60 years showed no differences in time to second therapy between groups defined by these age limits, while OS was significantly shorter for patients diagnosed when they were over the age of 50 or 60 years (Table 3).

Treatment toxicity

Patients tolerated the first therapy equally in the two cohorts. 32% of ≤40-year-olds and 26% of patients aged >40 years (P=0.25) required hospitalization and administration of intravenous antibiotics for neutropenic fever. The incidence of febrile neutropenia did not increase in the relapsed setting, affecting 29% and 38% (P=0.59) of young and old patients, respectively. There were no deaths due to bacterial sepsis following the administration of cytotoxic chemotherapy.

Clinical features at first relapse

Given the shorter period of first remission in young patients, we assessed the disease features seen at the time of second therapy (Supplementary Table 1). Complete blood count parameters were similar, except that the older patients had lower haemoglobin level at the time of second treatment (P=0.042). The indications for re-treatment were similar between the two groups. Cytopenia meeting the criteria for re-treatment was the most common indication present in 92% and 97% of young and old patients, respectively.

Secondary malignancies

The cumulative incidence of secondary malignancies in the 173 patients seen at MSKCC within 3 months of HCL diagnosis was compared in the young and old cohorts. No visceral malignancies or sarcomas were diagnosed in patients aged ≤40 years during the follow-up period, while such tumours accounted for 16/25 secondary cancers in the >40 years age group. All secondary cancers in those ≤40 years old were skin malignancies. The 10-year cumulative incidence of all skin cancers in patients ≤40 years old was 21% compared to 10% in patients >40 years of age (P=0.71) (Figure 1.4). Overall, the cumulative incidence of secondary malignancies, including non-melanoma skin cancers, was not significantly different between the two groups, with 10-year CIR of 21% and 29% in the young and old cohorts, respectively (P=0.22) (Figure 1.3). We assessed the cumulative impact of chemotherapy exposure on the incidence of secondary cancer in the two groups and found that the proportion of patients with secondary cancers in the older group increased progressively with each subsequent cycle of therapy, from 12%, 15% and 42% following 0, 1 or ≥2 lines of treatment (P=0.018).

The assessment of secondary cancer was extended to the entire cohort of 331 patients in order to describe the spectrum of secondary malignancies (Supplementary Table 4). Secondary cancers, including all skin cancer types, were diagnosed in 14/63 (22%) of young and 56/268 (21%) of older patients. The median time to any secondary cancer from the time of HCL diagnosis was 19 years in the ≤40 years age group and 4 years in the older cohort.

The most frequent cancers were non-melanoma skin cancers and melanoma in both age cohorts. Skin cancers made up 33% and 36% of all secondary malignancies found in the young and old age groups, respectively. There were two cases of therapy-related myeloid neoplasm in the ≤40 years age group and only 1 in the > 40 years age group, suggesting that the leukemogenic potential of cladribine and pentostatin is small. The first patient developed acute myeloid leukaemia (AML) following 4 cycles of cladribine, while the other two received fludarabine as a component of prior therapy. The single long-term therapy-related AML survivor is a 31-year-old man who underwent allogeneic stem cell transplantation and remains in remission from both AML and HCL 5 years post allograft. There were 7 cases of lymphoproliferative disorders other than HCL, and all occurred in the > 40 years age cohort. In the ≤ 40-year-old age group, 3/14 (21%) identified secondary cancers were visceral malignancies compared to 26/67 (39%) of all second cancers in the older cohort (P=0.36). The diversity of non-skin cancers was higher in those aged >40 years compared to younger patients.

Discussion

This study is the first to directly compare treatment outcomes and disease complications in young and old patients with HCL. The key novel findings of our analysis suggested that patients diagnosed with HCL before the age of 40 years have significantly shorter response following standard therapy and, as a result, require more individual lines of treatment to maintain disease control during the follow-up period. Despite more chemotherapy exposure, younger patients had a lower CIR of secondary malignancy compared to older patients, in whom the incidence of secondary cancers increased following subsequent cycles of chemotherapy.

The cohort of younger HCL patients described here had similar treatment response outcomes to those published previously (Piro, et al 1990, Rosenberg, et al 2014). We applied an age cut-off of 40 years to define young patients as this age cut-off allowed comparison of our results with those published in a previous report that applied the same age cut-off to describe the outcomes of young patients (Rosenberg, et al 2014). The cCR rate following first therapy in patients aged ≤40 years was similar to the rate described by Rosenberg et al (2014) following first treatment with cladribine and comparable to that described for all HCL patients (Goodman, et al 2003). The duration of first response appeared significantly longer in our cohort (120 months) compared to that described in the Scripps Clinic report (57 months; Goodman et al 2003). This may be because 52% of younger patients reported by Goodman et al (2003) had received other therapy prior to cladribine, resulting in shorter subsequent responses. The equivalent time described in this report is the true TNT calculated following first therapy. This appears longer than the median duration of first remission following cladribine (Goodman, et al 2003), most probably because we measured TNT as the absolute time between successive therapies, rather than time to disease relapse at which point most patients do not require treatment. The median second TNT in our series was not reached, and in the young HCL patients from the Scripps Clinic series (Goodman et al 2003) it was 27 months. This is an important finding given that younger patients had a TNT that was half that of older patients (82 months). The addition of rituximab to cladribine may prolong TNT compared with cladribine alone (Ravandi, et al 2011) and thus could be used as method of prolonging remission duration in younger individuals. There were fewer total deaths in patients ≤40 years of age in our cohort compared to that reported by Rosenberg (4/63 vs 13/88), which may be explained by the shorter median follow-up in our study; however, deaths due to HCL appear similar, with 2 in the cohort reported by Rosenberg et al (2014) and 1 in this report.

As expected, we found a trend to shorter median OS in older patients, which was accounted for by deaths from secondary malignancies. Sensitivity analysis using age cut-offs of 50 and 60 years found that survival was significantly shorter in older patients diagnosed above these age limits (Supplementary Table 3). The median OS was similar to that described by previous authors (Goodman, et al 2003, Rosenberg, et al 2014). Although the absolute difference was small, younger individuals received initial therapy sooner after diagnosis than older patients and had shorter periods of remission after first and second therapy and, as a result, required more individual lines of treatment during a similar follow-up period to older patients. A similar trend was identified in the Scripps clinic evaluation where the duration of remission in young patients following first cladribine therapy was 57 months compared with a median duration of remission of 98 months in a historical cohort of all HCL patients not restricted by age (Goodman, et al 2003). The reasons for this disparity are unclear but may be a result of more MRD negative CR or a longer cladribine half-life due to reduced renal clearance in older patients. Alternatively, underlying differences in disease biology may explain the more aggressive disease tempo in younger individuals. Molecular differences between HCL in young versus elderly patients are not currently known. Although we found no differences in the incidence of the BRAF V600E mutation and cytogenetics, the incidence of IGHV4-34 expression in these two groups would be relevant to assess given this may predict poor responses to therapy (Arons, et al 2009). It would also be relevant to compare the rate of MRD negative remission in young and old patients, as this may help explain the differences in response duration.

Published studies have shown a significant trend to increased rates of secondary malignancies in patients treated with cladribine, ranging between 1.6- and 2-fold higher than those of the general population, although no study has stratified these patients by age at HCL diagnosis (Goodman, et al 2003, Saven, et al 1998). We found that treatment of HCL using modern therapy is likely to contribute to secondary malignancies in older individuals where the incidence increased significantly as patients received more lines of cytotoxic therapy. Younger individuals appeared to develop secondary cancers after a longer time period from when HCL was diagnosed. While we cannot definitively conclude whether the cumulative incidence of secondary cancer increased as a result of exposure to more treatment or simply due to a longer follow-up, our results suggest that both patient age and cumulative treatment are likely to be important risk factors for secondary cancer development in classical HCL. Death from secondary cancer was the most common cause of mortality, which is in-line with previous reports (Kurzrock, et al 1997, Rosenberg, et al 2014).

In conclusion, patients diagnosed prior to the age of 40 years account for 20% of patients with HCL. Young and old patients have similar clinical and pathology features at diagnosis and at the time of relapse. Older individuals showed a trend to higher incidence of secondary malignancies and a significant increase in the incidence of secondary cancers with more chemotherapy exposure. Both groups had very durable remissions with prolonged survival when treated with purine analogues. Patients diagnosed prior to the age of 40 years had a period of first remission that was half of that seen in patients aged >40 years, equating to an absolute difference of 82 months between 1^st and 2^nd treatments. Assessing MRD following first therapy in younger patients and using rituximab following purine analogues in those with detectable residual disease is likely to lengthen the remission duration in younger patients and will reduce exposure to potentially mitogenic chemotherapy.