A matched comparison of cyclophosphamide, bortezomib and dexamethasone (CVD) versus risk-adapted cyclophosphamide, thalidomide and dexamethasone (CTD) in AL amyloidosis

Abstract

Despite improvements in therapy amyloid light-chain (AL) amyloidosis, there are few studies comparing different regimens. Here we present a matched comparison with 69 patients in each cohort examining upfront therapy with cyclophosphamide, bortezomib and dexamethasone (CVD) vs cyclophosphamide, thalidomide and dexamethasone (CTD). On an intention-to-treat basis, the overall response rates were 71.0% vs 79.7% in the CVD and CTD arms, respectively, (P=0.32). A higher complete response (CR) rate was observed in the CVD arm (40.5%) vs CTD (24.6%), P=0.046. One-year overall survival (OS) was 65.2% and 66.7% for CVD and CTD, respectively (P=0.87). The median progression-free survival (PFS) was 28.0 and 14.0 m for CVD and CTD, respectively (P=0.039). In a landmark analysis assessing outcomes performed at 6 months, the CR rate with CVD was 59.6% vs 34.0% for CTD (P=0.03). The 1-year OS was 96% with CVD and 92% with CTD (P=0.40). The median PFS with CVD was not reached and was 19.2 m with CTD, P=0.028). In summary, both regimens are unable to overcome the high rate of early deaths in AL amyloidosis. However, CVD correlates with improved depth of response and superior PFS supporting its use in the frontline setting. Further optimisation and better supportive-care strategies are required to increase the proportion of patients fully benefiting from therapy.

Introduction

Amyloid light-chain (AL) amyloidosis is a disease of protein fibril deposition most commonly associated with an underlying clonal plasma cell dyscrasia. It often complicates multiple myeloma, but is also seen in low-grade lymphomas. The precursor protein of the amyloid fibrils in AL amyloidosis are the abnormal light chains produced by the underlying clone. The clinical presentation is heterogeneous depending not only on the underlying clonal disorder but also the spectrum and severity of organ involvement.¹ The poor survival associated with AL amyloidosis is largely driven by outcomes in patients with advanced cardiac disease.^{2, 3, 4}

At present, only two phase III trials have ever been conducted in this disease and another is ongoing. The first established alkylators as a cornerstone of therapy in this disease.⁵ The second compared high-dose melphalan and autologous stem cell transplantation with low-dose oral melphalan.⁶ Reported outcomes vary considerably based on the severity of organ involvement within a given cohort.^{6, 7, 8} Recent data presented by our own group as well as others suggests further improvements in outcomes since the introduction of novel agents.^{2,9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20} Thalidomide-based combinations showed promise, given the rapidity of action with clonal responses seen within 1–2 cycles of therapy but with significant toxicity.^{21, 22, 23} Of particular interest has been the recent success using proteosome inhibition, given the increased sensitivity of the underlying problematic plasma cells due to the proteotoxicity of misfolded light chains.^24,25 Recent retrospective^18,19,26,27 and prospective studies^13,17,28 suggest that regimens incorporating a proteosome inhibitor can achieve rapid and deep clonal control with the hope that this may improve survival and organ function over the long term.

Up to 50% of high-risk patients may die within the first 6 months contributing to the early deaths seen in many of the currently published series.^{2, 3, 4} Treatment options for such patients remain uncertain. Proteasome inhibitors are attractive owing to the rapidity of haematologic response, as well as data showing cardiac responses in patients achieving deep clonal control.^{17, 18, 19}

Although encouraging, there is little prospective evidence available, nor any comparative studies with current standards of care to formally assess the efficacy of proteasome inhibition in this disease, especially in the upfront setting. There has been a general perception that proteasome inhibitor-based therapy is ‘superior’ to standard treatments. Oral cyclophosphamide, thalidomide and dexamethasone (CTD) has been the standard of care in the United Kingdom over the last decade, but increasing number of patients are treated using frontline bortezomib combinations. This has also lead to closure of the North American arm of a randomised phase III trial, run by the Eastern Cooperative Oncology Group, designed to assess the impact of adding bortezomib to melphalan–dexamethasone due to lack of recruitment. An international collaborative study continues in Europe and Australia.

Although prospective data are awaited, here we present a matched comparison of patients treated upfront with cyclophosphamide, bortezomib and dexamethasone (CVD) or CTD, examining response and survival end points. We hypothesised that an increased depth of response would be observed in the CVD arm translating into improvements in progression-free survival (PFS) and overall survival (OS), and may overcome early deaths in patients with advanced cardiac involvement.

Patients and methods

Patients

The primary cohort is comprised of 138 matched patients with systemic AL amyloidosis—69 in each arm treated with frontline CVD or risk-adapted CTD, respectively—assessed at the National Amyloidosis Centre in London between July 2008 and July 2012. Amyloidosis of the AL subtype was defined by the presence of Congo red-positive fibril deposition on biopsy, and immunohistochemistry or mass spectrometry was used to prove either kappa or lambda light-chain specificity of the fibrils. Evidence of clonality was required (monoclonal protein in the serum or urine, light-chain excess on serum-free light-chain analysis or clonal population of plasma cells within the marrow). Patients were excluded if they had a non-plasma cell-based clonal disorder giving rise to the amyloidogenic protein or if they did not have serologic markers present to follow response. Organ involvement and Mayo staging was determined as per published criteria.^{29, 30, 31, 32}

The CVD cohort reflected all patients treated with this regimen in the upfront setting seen at the National Amyloidosis Centre until July 2012. Patients were then matched with a cohort of patients treated with CTD as first-line therapy. Patients were matched on a one-to-one basis with CVD used as the base cohort and serial patients selected from a total of 181 CTD-treated patients assessed over the same time period. Patients were matched by Mayo stage. We also attempted to have similar proportions of patients with a high difference in free light chains (dFLC) - defined as the difference between the involved and uninvolved light chain being < or ⩾ 180 mg/l as well as a very high NT-proBNP (NT-proBNP < or ⩾ 8500 pmol/l); ^2,3,33 both factors shown to be prognostic in AL amyloidosis.

The CVD and CTD regimens were delivered as previously described.^18,20 Dose modifications were at the discretion of the treating haematologist. Owing to the importance of achieving a rapid clonal response in AL amyloidosis, a strategy of an early switch to an alternative regimen is routinely followed in the United Kingdom for those patients achieving <90% reduction in the dFLC after three cycles of therapy.³⁴ Irrespective of the regime chosen at time of early switching, patients were analysed based on the regimen used at the outset of treatment.

Haematologic and organ responses were determined as per recently published criteria.^31,32 Both conventional haematologic responses and dFLC responses were examined, and the maximal response achieved post-frontline therapy prior to relapse was recorded. dFLC-very good partial response (VGPR) was defined as attaining a dFLC of <40 mg/l post therapy. Patients dying before a response could be documented were deemed non-responders. Haematologic and organ responses were recorded as the best response achieved prior to clonal relapse. For the intent to treat (ITT) analysis, in patients who underwent an early switch in therapy due to a sub-optimal response, the maximal response was recorded as the best response achieved even if it occurred post switch, provided the patient did not meet criteria for clonal relapse before the new regimen was initiated. Pre-switch response rates were reported separately as the best response achieved before requiring a switch.

Statistical analysis was performed using SPSS version 20. Data were analysed on a ITT basis. For qualitative variables descriptive analyses were based on frequencies. Frequencies were compared using χ²-square or Fischer’s exact test. Quantitative variables were analysed by means, s.d. and medians. Comparisons were made using the Mann–Whitney U-test. Response outcomes by treatment group were compared using two-tailed Pearson and Spearman correlative statistics. A P-value <0.05 was considered significant.

Survival end points were calculated on an ITT basis by the Kaplan–Meier method and the use of the log-rank test and univariate Cox proportional-hazards regression analysis. Median OS was calculated from the start of treatment until death or last follow-up. PFS was examined in patients achieving a partial response or better and was measured from the start of treatment until relapse, death or last follow-up. This was measured as the time from treatment initiation to haematologic progression. All P-values were two sided with a significance level of 0.05.

To examine the impact of each regimen on those that survived long enough to have potentially benefited from a complete course of therapy, a landmark analysis was performed in patients surviving at least 6 months from the start of treatment. All survival and response end points as above were examined in this cohort.

Results

Detailed patient characteristics are described in Table 1. The cohorts were matched based on Mayo stage and there were no significant differences in the baseline characteristics. Specifically the groups were balanced with respect to known prognostic factors identifying high-risk AL amyloidosis (Mayo stage III by conventional classification, dFLC>180 mg/l, blood pressure and NT-proBNP>8500 ng/l).

Table 1 Patient characteristics

On an ITT basis, there was no difference in the median number of cycles received in each group (4.0 and 4.0 for the CVD and CTD arms, respectively, P=0.82). In addition, a similar proportion of patients received four or more cycles of therapy (50.7% vs 59.4%, P=0.31). In the CVD and CTD cohort, respectively, 1 (1.4%) and 14 (20.3%) switched therapies after cycle 3 owing to lack of achieving >90% dFLC response (P<0.005). For the CTD patients, 13 received a bortezomib-containing regimen and 1 a lenalidomide-containing regimen. On an ITT basis, overall haematological response rates were comparable (49 (71.0%) vs 55 (79.7%)) in the CVD vs CTD arms, respectively (P=0.32). There was, however, a significantly higher complete response (CR) rate in the CVD vs CTD arm (28 (40.6%) vs 17 (24.6%), respectively, P=0.046, Figure 1). The pre-switch CR was achieved in 28 (40.6%) and 13 (18.8%) patients in the CVD and CTD arms, respectively, (P=0.002). Thirty-five (50.7%) and 31 (44.9%) patients achieved ⩾VGPR in the CVD and CTD arms, respectively, (P=0.5). Of those who switched from CTD, 8 (57.1%) improved their response beyond a partial response (4 (28.6%)—CR, 4 (28.6%)—VGPR). For the single patient in the CVD arm, thalidomide was added to the regimen due to a <50% dFLC response with the patient subsequently attaining a partial response. There was no significant difference in response rates and survival for patients over age 65 years in either cohort, but small numbers limit interpretation.

Figure 1

Response rates for patients receiving upfront CVD or CTD. Response rates are listed by regimen for the ITT, 3-month pre-switch (this group excluded patients switched to an alternative regime after 3 cycles of therapy) and LM cohorts. Similar overall responses are seen between the groups, but statistically significant improvements in CR rates are observed in favour of CVD for both the ITT, pre-switch and landmark cohorts. ITT=intention-to-treat, LM=landmark.

Organ responses were seen in both cohorts and are described in Table 2. We examined organ responses on an ITT basis and separately in the patients who did not switch therapy. Response assessment required baseline involvement of the organ in question, as well as for the patient to be alive long enough to assess response. Thirty-three and 26 patients in the CVD and CTD cohorts, respectively, were eligible for cardiac response assessment. Fifteen and 13 patients were eligible for hepatic response. Thirty and 41 were eligible for renal responses.

Table 2 Organ responses: expressed as percentage of patients with baseline organ involvement who were alive and assessable for organ response

The median follow-up was 12.7 and 25.5 m for the CVD and CTD arms, respectively. The median OS has not been reached in either group, and there was no statistically significant difference in the 1-year OS (65.2% vs 66.7% for CVD and CTD, respectively, P=0.87, Figure 2a). Patients treated with CVD had a significantly prolonged PFS, twice as long as those treated with CTD (28.0 vs 14.0 m, P=0.039, Figure 2b).

Figure 2

Survival end points in the ITT cohort comparing upfront CVD (solid line) with CTD (dashed line). OS remains similar in the ITT population (a). There is a statistically significant improvement in PFS in favour of those treated with CVD (b).

For responding patients, the median time to first and maximal response was not significantly different for the CVD and CTD arms (1.2 vs 1.0 m (P=0.22) and 2.2 vs 2.8 m (P=0.25), respectively). Eleven (15.9%) of CVD and 7 (13.1%) of CTD patients died within 6 weeks (P=0.24). Twenty-two (31.9%) CVD and 19 (27.5%) CTD patients died within 6 months (P=0.22). This phenomenon was restricted to the high-risk patients, as 100% of the early deaths in each arm had Mayo stage III disease. In both the CTD and CVD groups there was a substantial population of patients who failed to achieve maximal response owing to early deaths. NT-proBNP>8500 ng/l identified those at very high risk of early death. Neither regimen correlated with improved survival irrespective of the NT-proBNP levels (Figure 3).

Figure 3

OS for patients treated with CVD (solid line) and CTD (dashed line) based on NT-proBNP status. There is no significant difference in survival between treatment groups based on cardiac risk by NT-proBNP above (a) or below (b) 8500 ng/l.

In the landmark analysis, examining patients surviving beyond 6m, there were 47 patients in the CVD arm and 50 in the CTD arm. The overall haematologic response rate remained similar and the correlation between higher CR rates and frontline CVD persisted (28 (59.6%) vs 17 (34.0%), respectively, P=0.03, Figure 1). This is despite the 18.8% of patients in the CTD arm receiving a bortezomib combination for inadequate response. The OS was similar with a 1-year OS of 96% in the CVD arm and 92% in the CTD arm (P=0.40, Figure 4a). The median PFS was statistically superior with the CVD arm compared with CTD (median not reached (59% at 2 years) vs 19.2 m, P=0.028, Figure 4b). This may be partly driven by the increased CR rate in the CVD cohort compared with the CTD arm in the landmark population. There was no difference in the median number of cycles delivered in the landmark cohort (4.0 and 4.0 for the CVD and CTD arms, respectively, P=0.92) or percentage of patients receiving four or more cycles of therapy (25.6% vs 26.0%, P=0.68).

Figure 4

Survival end points in the landmark analysis at 6 months post treatment initiation comparing upfront CVD (solid line) with CTD (dashed line). OS remains similar in the landmark population (a). There is a statistically significant improvement in PFS for patients treated with CVD (b).

Discussion

The treatment of AL amyloidosis has improved with the more widespread use of novel agents. Oral melphalan and dexamethasone is often considered the ‘standard of care’ for treatment of AL amyloidosis, but median time to response is 4 months.^7,8 Over the last decade, in the United Kingdom, CTD has been the standard frontline treatment for AL amyloidosis owing to the potential for more rapid response rates;^20,21 an important feature especially in high-risk patients where rapid clonal control is thought to be integral to preserving organ function and improving survival.³⁵ However, the use of CTD in this disease is often a challenge owing to toxicities^20,21,23 leading to frequent dose adjustments and discontinuations potentially limiting the ability of patients to fully benefit from treatment.³⁶ Given its particularly potent activity against cells with high protein stress,^24,25 it was with great interest that the use of bortezomib was investigated in AL amyloidosis. A number of early single-institution series^26,27 lead to a phase I/II study in relapsed refractory patients,¹⁷ showing unprecedented depth of response. Spurred on by the success of triplet-based combinations with an alkylator/steroid backbone, regimens such as CVD have since shown even higher, deeper and more rapid haematologic responses especially in the frontline setting.^18,19

Despite these advances in therapy, no head-to-head trials have been performed to examine the optimal novel agent-containing regimen or indeed compare a novel agent-based regimen with a traditional alkylator-based therapy. Nor has it been prospectively proven that a rapid clonal response will improve outcomes in the poorest risk sub-group of patients. In AL amyloidosis, one is striving for rapid control of the underlying clonal disease, but also contending with significant treatment-related morbidity exacerbated by the fibril-related organ dysfunction. Studies in advanced AL are difficult given the relative rarity of the disease; the high mortality rate, which often exceeds 50% within the first 6 months, adds further challenges and marked reluctance from funding organisations as well as commercial sponsors for undertaking clinical trials in this setting.^{2, 3, 4,6}

In the current series, on an ITT basis, although there was no statistically significant difference in overall haematologic response rates between the groups, the CR rate associated with CVD was significantly greater suggesting superior clonal control with this regimen in the upfront setting. In responding patients, this translated into a superior PFS in favour of CVD. This is particularly notable as bortezomib was used in 20% of patients in the CTD arm as part of the described ‘switch protocol’ emphasizing the importance of early use of a proteasome inhibitor. In the United Kingdom, a strategy of rapid treatment-switching has been adopted for patients who fail to achieve a dFLC-VGPR within the first three cycles of upfront therapy.³⁴ For those treated with CTD upfront this is most often to a bortezomib-containing regimen, but the final decision to switch as well as the specific agent used is influenced by toxicity, organ involvement, patient preference and regional drug availability. It was striking that in the CTD arm of this study, 20% of patients went on to switch therapy due to lack of ‘adequate’ early response compared with only one patient in the CVD arm. This ‘cross-over’ effect likely contributes to the lack of significant differences in OS as well as rates of VGPR. That said, only 57.1% of patients who switched therapy in the CTD arm improved their responses beyond a partial response. However, on an ITT basis, there is still a statistically significant benefit in terms of CR rates and PFS in the CVD arm—suggesting that use of the proteosome inhibitor at the outset of therapy has greater clonal benefit, and sub-optimal upfront therapy may allow for development of resistant clones.

Analysis of the landmark cohort further supports this finding, as in this population ~60% of those receiving CVD attained a CR again correlating with a longer PFS. Examining patients surviving long enough to potentially maximally benefit from treatment is a particularly important consideration in therapeutic trials for AL amyloidosis, especially in the frontline setting. Failing to recognize the impact of early deaths, many of whom are recorded as ‘non-responders’, may underestimate the true impact of a given regimen on surviving patients. In the landmark cohort in this series, the CR rates are similar to previous studies with this regimen that contained fewer Mayo stage III patients in the upfront setting, where the early-death rate was lower than that reported here.^18,19

It is interesting to note that the proportion of patients achieving a dFLC-VGPR or better was similar in each arm. Although achieving dFLC-VGPR is now considered an acceptable goal of therapy in AL amyloidosis, a large multi-centre series still shows a statistically significant improvement in survival between those achieving a CR versus those who attain a VGPR.³² This reinforces the fact that a CR remains the optimal goal of any therapeutic intervention in this disease.

In both the ITT as well as the landmark analysis, the mean number of cycles was similar between groups, and half the patients in each arm received four or more cycles of treatment. This suggests that any differences in outcome were not due to insufficient duration on therapy in the CTD arm, although the retrospective nature of the series limits ability to calculate actual doses of drugs received.

Although the death rate of ~30% within the first 6 months in each arm is consistent with the natural history of this disease using conventional therapies,³⁶ it is frustrating to note that neither CVD nor CTD was associated with a reduction in early deaths. This is despite the fact that clonal responses could be documented in most patients within the first 1–2 cycles. The high proportion of Mayo stage III patients (~20% of whom had an NT-proBNP >8500 ng/l) likely accounts for this observation, and contributed to the lower than expected response rates compared with those recently reported with upfront CVD.^18,19 This study clearly shows that early, rapid and deep responses are unable to overcome the disease’s natural history in advanced cardiac AL amyloid. In fact, there was a slight but non-significant excess of early deaths in the CVD arm. The reason for this remains uncertain, but CVD patients receive therapy in hospital and are more likely to have received full doses prescribed compared with home-based oral chemotherapy, where inadvertent dose modification is very common. Further analysis and accrual of data is ongoing to investigate this in more detail.

Unfortunately, owing to the small numbers of patients within each Mayo risk group an analysis by stage could not be performed. Neither regimen was able to overcome the extremely poor outcome in patients with very high-risk disease as defined by NT-proBNP>8500 ng/l. This should be an important consideration in future studies, given the markedly different expected OS outcomes in Mayo stage III vs non-Mayo stage III patients, as well as those with high-risk Mayo stage III disease. It is likely that a risk-adapted approach, balancing toxicity with clonal control and emphasizing supportive-care strategies in these sick patients, will be required to overcome the early deaths.

Clonal control is of tantamount importance in AL amyloid, as the light chains themselves are the disease. Limiting the production of the amyloidogenic free light chains is an absolute requirement for organ responses to occur and is the only way to prevent further deterioration. In this series, organ responses are seen and are similar in both treatment arms even when examining only those who did not switch therapy. Although the trend favouring CVD is encouraging, especially with respect to cardiac responses, the differences are not statistically significant. However, follow-up is short, especially in the CVD cohort, and may be further compounded by the potential cross-over effect from the switch protocol. The contribution of immunomodulator-based therapy-related increase in cardiac biomarkers remains a potential confounder in the CTD arm for biomarker-based assessment of cardiac responses. It can take many years for marked improvements to be seen with respect to the fibril load.³⁵ In addition, based on recent series, the expected OS in those who live through and respond to therapy can be as long as a decade^37,38 even in the relapsed setting.³⁹ Depth of clonal response has been shown to be critical in organ function over the long term,^35,40 and there is a substantial increase in the proportion of patients attaining the optimal CR end point after CVD exposure. Provided the circulating precursor protein levels remain below the threshold required for amyloid fibril formation, one would expect ongoing improvements in total body amyloid load with longer follow-up. It is important to note that current criteria for cardiac response examines NT-proBNP levels at 6 months post therapy.^31,32,41 This is the time point that was shown to be prognostic for OS in the initial publication. Although this provides a means to rapidly assess improvements in cardiac outcomes, it does not take into account the fact that many patients will have ongoing improvements in cardiac biomarkers well beyond the 1-year mark.

The results of this study should be interpreted with caution, given the retrospective nature of the data. That said, comparative trials in AL amyloid are difficult to perform. The results presented here are encouraging and are consistent with recently published retrospective and prospective series using bortezomib-based regimens.^{17, 18, 19,27,42} In summary, upfront therapy with CVD correlates with deeper response rates and improved clonal control. This results in improved PFS in responding patients in comparison with risk-adapted CTD by both ITT and landmark analysis. Neither regimen was able to overcome the expected early deaths seen in this disease, despite rapid clonal control. Further optimisation and better supportive-care strategies are required during the early stages of treatment to increase the proportion of patients who will live long enough to maximally benefit from therapy. Longer follow-up is required to assess the impact on OS and long-term organ function. Ongoing phase III trials are currently underway to address these issues in a prospective manner.