Impact of cytogenetic abnormalities on treatment outcomes in patients with amyloid light-chain amyloidosis: subanalyses from the ANDROMEDA study

Abstract Background Cytogenetic abnormalities are common in patients with amyloid light-chain (AL) amyloidosis; some are associated with poorer outcomes. This post hoc analysis of ANDROMEDA evaluated the impact of certain cytogenetic abnormalities on outcomes in this patient population. Methods Patients with newly diagnosed AL amyloidosis were randomised 1:1 to daratumumab, bortezomib, cyclophosphamide, and dexamethasone (D-VCd) or VCd. Outcomes were evaluated in the intent-to-treat (ITT) population and in patients with t(11;14), amp1q21, del13q14, and del17p13. Results Overall, 321 patients had cytogenetic testing (D-VCd, n = 155; VCd, n = 166); most common abnormalities were t(11;14) and amp1q21. At a median follow-up of 20.3 months, haematologic complete response rates were higher with D-VCd vs VCd across all cytogenetic subgroups and organ response rates were numerically higher with D-VCd vs VCd across most subgroups. Point estimates for hazard ratio of major organ deterioration-PFS and -EFS favoured D-VCd over VCd for all cytogenetic subgroups. Deep haematologic responses (involved minus uninvolved free light chains [FLC] <10 mg/L or involved FLC ≤20 mg/L) were seen in more patients with D-VCd than VCd in all ITT and t(11;14) cohorts. Conclusions These results support the use of D-VCd as standard of care in patients with newly diagnosed AL amyloidosis regardless of cytogenetic abnormalities.


Introduction
Amyloid light-chain (AL) amyloidosis is a plasma cell disease characterised by the production of light chains that form amyloid fibril deposits in tissues leading to organ dysfunction and death [1]. Like multiple myeloma (MM), AL amyloidosis is associated with a high frequency of cytogenetic abnormalities, with 60-90% of patients with AL amyloidosis having at least one chromosome abnormality [2]. Although MM and AL amyloidosis do share a similar spectrum of cytogenetic abnormalities, the distribution of these aberrations varies between the diseases.
In patients with AL amyloidosis, t (11;14) is the most common abnormality, occurring in 40-60% of patients [2], compared with 15% of patients with MM [3]. The prognostic value of cytogenetic abnormalities is well established in MM [4]. In particular, the presence of t(4; 14), del(17p), and 1q21 gain are indicative of unfavourable responses [5,6]. While the use of cytogenetic testing in predicting outcomes for AL amyloidosis is still evolving [2,7], patients with t (11;14) have shown poorer responses and overall survival when treated with bortezomibbased regimens compared to patients without t (11;14) [8,9].
Results from the phase 3 ANDROMEDA study (NCT03201965) demonstrated that the addition of daratumumab to bortezomib, cyclophosphamide, and dexamethasone (D-VCd) was associated with higher rates of haematologic complete response (CR), cardiac and renal responses at 6 months, and major organ deterioration-progression-free survival (PFS) compared with VCd alone [10]. Based on these results and an acceptable safety profile, D-VCd became the first therapy to be approved for the treatment of AL amyloidosis [10,11].
In this post hoc analysis of the ANDROMEDA study, we evaluated the distribution of cytogenetic abnormalities in patients with AL amyloidosis and assessed whether the benefit of D-VCd compared with VCd in AL amyloidosis was maintained in patients with cytogenetic abnormalities.

Patients and design
A primary report of the ANDROMEDA study has been previously published [10]. In brief, ANDROMEDA is a randomised, open-label, active-controlled, multicenter, phase 3 study in patients with newly diagnosed AL amyloidosis. Key eligibility criteria included newly diagnosed AL amyloidosis with measurable haematologic disease, !1 involved organ, cardiac stage I-IIIa (based on the Mayo clinical staging system [12,13]), estimated glomerular filtration rate !20 ml/min/1.73 m 2 , and no prior diagnosis of symptomatic MM. Patients with evidence of significant cardiovascular conditions were excluded from the trial.
Patients (N ¼ 388) were randomised 1:1 to D-VCd or VCd for 6 cycles of 28 days; randomisation was stratified by cardiac stage (Stage I, II, and IIIa), countries that typically offer or do not offer transplant for patients with AL amyloidosis, and renal function (CrCl !60 ml/min or <60 ml/min). Thereafter, patients in the D-VCd group received daratumumab alone every 4 weeks until disease progression, the start of subsequent therapy, or for a maximum of 24 cycles from the start of the treatment, whichever occurred first. All patients received subcutaneous (SC) bortezomib (1.3 mg/m 2 weekly), oral or intravenous cyclophosphamide (300 mg/m 2 weekly), and oral dexamethasone (20-40 mg weekly). SC daratumumab (1800 mg co-formulated with recombinant human hyaluronidase PH20 in 15 ml) was administered by weekly injection in cycles 1-2 and every 2 weeks in cycles 3-6 and every 4 weeks thereafter until disease progression, the start of subsequent therapy, or for a maximum of 24 cycles from the start of the treatment, whichever occurred first.

Assessments
The proportion of patients with t(11;14), 1q21 amplification (amp1q21), del13q14, and del17p13 was calculated based on fluorescence in situ hybridisation (FISH) and/or karyotyping. Cytogenetic testing was performed locally and was not predefined in the study protocol. Patients were considered to have an abnormal gene if detected by either FISH or karyotyping.
Patients with more than one cytogenetic abnormality were also examined if they had !2 abnormalities from the four genes tested.
The primary endpoint of the ANDROMEDA study was rate of overall (at any time from randomisation to clinical cut-off date [CCO]) haematologic CR, which was defined as an involved free light-chain level less than the upper limit of the normal range with negative serum and urine immunofixation; normalisation of the uninvolved free light-chain level or free light-chain ratio was not required to determine a complete response. Haematologic CR was evaluated for the intentto-treat (ITT) population (i.e. all randomised patients) and those expressing t (11;14), amp1q21, del13q14, and del17p13. Haematologic responses were evaluated weekly for cycle 1, every 4 weeks for cycles 2-6, and every other month thereafter until major organ deterioration-PFS, death, withdrawal of consent to participate, or end of the study. Responses were adjudicated by an Independent Review Committee.
Secondary endpoints included major organ deterioration-PFS, major organ deterioration-event-free survival (EFS), and cardiac and renal response rates at 6 months. Major organ deterioration-PFS is a composite endpoint defined as end-stage cardiac disease (requiring cardiac transplant, left ventricular assist device, or intra-aortic balloon pump), endstage renal disease (requiring haemodialysis or renal transplant), haematologic progression per consensus guidelines [14], and death (whichever came first). For major organ deterioration-EFS, an event was defined as haematologic progression, major organ deterioration, initiation of subsequent non-cross resistant anti-plasma cell therapy, or death (whichever came first).
Cardiac and renal response rates were defined as the proportion of baseline organ-evaluable patients who achieved a response at 6 months. Cardiac response was based on N-terminal prohormone of brain natriuretic peptide (NT-proBNP) response (>30% and >300 ng/L decrease in patients with baseline NT-proBNP !650 ng/L) or New York Heart Association (NYHA) class response (>2 class decrease in patients with baseline NYHA class 3 or 4) per consensus criteria. Renal response was defined as !30% decrease in proteinuria or proteinuria decreased to <0.5 g/24 h in the absence of renal progression (50% increase [at least 1 g/day] of 24 h urine protein to >1 g/day or 25% worsening of serum creatinine or creatinine clearance), as developed by a group of international experts [15]. Endpoints were evaluated for all ITT,t(11;14), amp1q21, del13q14, and del17p13.

Statistical analysis
No statistical comparisons were performed between the cytogenetic subgroups. Descriptive statistics summarised baseline characteristics, number of subsequent lines of therapy, and responses (haematologic, cardiac, and renal responses) by cytogenetic subgroup and by baseline difference between involved and uninvolved serum free light chains (dFLC) and plasma cell percentage.
The proportion of patients with specific cytogenetic abnormalities was calculated from evaluable patients in each treatment group. For the calculation of the proportion of patients with a cytogenetic abnormality plus an additional abnormality, the numerator was the abnormality of the specified gene plus 1 additional chromosomal abnormality. The denominator was the total number of patients with the specified chromosome tested plus !1 additional gene tested using FISH and the number of patients who had whole bone marrow karyotyping performed.
The data for haematologic CR and organ response rates with D-VCd and VCd utilised the CCO of November 2020 (median follow-up, 20.3 months), and data for major organ deterioration-PFS and -EFS were based on the CCO of February 2020 (median follow-up, 11.4 months).
Odd ratios (ORs) and 95% confidence intervals (CIs) for haematologic CR rate and organ response comparing the two treatment groups across subgroups were estimated using unstratified Mantel-Haenszel estimation. Nominal p-values for the subgroups were calculated using Fisher's exact test. Regarding the ITT population, the ORs and 95% CIs for haematologic CR rate and organ response comparing the two treatment groups were obtained through Mantel-Haenszel estimate of the common OR for stratified tables based on the randomisation stratification factors. P-values were calculated by the Cochran-Mantel-Haenszel Chi-Squared test.
The hazard ratios (HRs) and 95% CIs for major organ deterioration-PFS and -EFS with D-VCd and VCd across cytogenetic subgroups were evaluated using a Cox proportional hazards model with treatment as the sole explanatory variable. The HR and 95% CI for major organ deterioration-PFS in the ITT population were obtained from unstratified weighted Cox proportional hazards model, including treatment group as the sole explanatory variable by using the Inverse Probability of Censoring Weighting method. The HR and 95% CI for major organ deterioration-EFS in the ITT population were obtained from a Cox proportional hazards model with treatment as the sole explanatory variable and stratified by randomisation stratification factors.
Additional analyses included the proportion of patients achieving a deep haematologic response defined as the dFLC <10 mg/L or iFLC 20mg/L. Further analyses evaluated the impact of the presence or absence of t (11;14) and amp1q21 on depth of response by treatment as measured by the proportion of patients achieving haematologic CR and very good partial response or better (!VGPR). Nominal p-values were calculated using Fisher's exact test.

Baseline characteristics and demographics
Baseline characteristics were generally similar between subgroups of patients based on cytogenetic abnormalities (Table 2) and the entire ITT population. Median age across subgroups was approximately 65 years. The proportion of males varied across the cytogenetic subgroups, with fewer in the del17p13 cohort (38.9%) and more in the t(11;14) cohort (72.7%). Across the 4 cytogenetic subgroups, the proportion of patients with cardiac and renal involvement was generally consistent, and 41.3-55.5% of patients had Mayo Cardiac Stage III disease. Patients had a median of two involved organs across all cytogenetic abnormalities.

Haematologic CR rate
At a median follow-up of 20.3 months, the haematologic CR rate was higher with D-VCd vs VCd in all cytogenetic subgroups, including patients with 2 or more cytogenetic abnormalities, and were generally comparable to responses in the entire study population (Figure 2(A), Supplemental Figure 1) and to the primary report at 11.4 months (D-VCd, 53.3% and VCd, 18.1%) [10]. Haematologic CR at 6 months was higher with D-VCd vs VCd in all cytogenetic subgroups, in patients with 2 or more cytogenetic abnormalities, and the ITT population (Figure 2(B)).
Time to first response (partial response or better) was shorter for patients receiving D-VCd than VCd and similar between patients in the overall population (median [range]; D-VCd, 11

Organ response
Similarly, cardiac and renal responses at 6 months were numerically higher with D-VCd vs VCd across all cytogenetic subgroups as well as in patients with 2 or more cytogenetic abnormalities (Figure 3), with the exception of the cardiac response in the del17p13 cohort where D-VCd was slightly lower than VCd. Results were comparable to responses seen in the ITT population.

Major organ deterioration-PFS and -EFS
In the entire ITT population, median major organ deterioration-PFS was not reached in either treatment group. This was also seen in 3 of the 4 subgroups, with only the VCd  arm of the del17p13 group reaching median major organ deterioration-PFS (7.5 months). Point estimates for betweengroup comparisons favoured D-VCd for all subgroups, with HRs ranging from 0.18 to 0.89 (Figure 4), although the CIs were wide and crossed unity for all cytogenetic subgroups. For the ITT population and each of the cytogenetic subgroups, median major organ deterioration-EFS was not reached in patients treated with D-VCd but was reached in patients treated with VCd in the ITT population (8.8 months) and across all cytogenetic subgroups (range, 7.0 months to 13.4 months). As with major organ deterioration-PFS, point estimates of the HR for major organ deterioration-EFS favoured D-VCd over VCd for all cytogenetic subgroups, ranging from 0.23 to 0.53 ( Figure 5). However, like the results for the major organ deterioration-PFS, the CIs were wide, and some (del17p13 and amp1q21) crossed unity.

Discontinuations and subsequent lines of therapy
Fewer patients discontinued study treatment or received subsequent therapy in the D-VCd vs VCd treatment arms in the entire ITT population and with each cytogenetic abnormality (Table 3).
Across all cytogenetic subgroups and in the entire ITT population, patients in the D-VCd group went on to receive 1 or more subsequent lines of therapy less frequently than patients in the VCd group (Table 4). Overall, other antineoplastic agents (not including alkylating agents) were received as subsequent therapy by 37.8% of patients in the VCd treatment arm, most commonly daratumumab and bortezomib. The use of daratumumab as a subsequent therapy was numerically highest in VCd-treated patients with amp1q21 (50.0%). Of the patients in the D-VCd treatment arm, 3.6% went on to receive subsequent anti-neoplastic treatment, most frequently daratumumab retreatment followed by bortezomib and venetoclax. Impact of the presence or absence of t(11;14) and amp1q21 on depth of response by treatment group t(11; 14) In the D-VCd group, rates of haematologic CR and ! VGPR were unaffected by the presence or absence of t(11;14) ( Table 5). However, in the VCd group, rates of haematologic CR and ! VGPR were numerically lower in patients with than without t(11;14), although not statistically significant.

amp1q21
In the D-VCd group, rates of haematologic CR and ! VGPR were comparable in patients with and without amp1q21 ( Table 5). In the VCd group, the rate of haematologic CR was numerically lower in patients with than without amp1q21, while the rate of ! VGPR was unaffected by the presence or absence of this cytogenetic abnormality.

Proportion of patients achieving a deep response
The proportion of patients achieving a deep response (either a dFLC <10 mg/dL or iFLC 20 mg/L) was higher in those patients receiving D-VCd vs VCd in the ITT population. In the t(11;14) subgroup, the proportion of patients achieving a deep response was similar to the ITT population with D-VCd but lower with VCd ( Figure 6). Impact of degree of bone marrow involvement at baseline on haematologic complete response and organ response rates

Impact of baseline dFLC on response rates
Overall, in patients who received D-VCd, both haematologic CR and cardiac response rates were unaffected by baseline dFLC (<18 mg/dL compared with !18 mg/dL, Table 6), while renal responses were numerically lower in patients with dFLC !18 mg/dL. In patients who received VCd, haematologic CR was numerically lower in those with a baseline dFLC !18 mg/dL in the ITT population, while cardiac and renal responses in the ITT population were unaffected by baseline dFLC.

Impact of baseline plasma cell percentage on response rates
In both treatment arms of the ITT population, haematologic CR was lower in patients with a baseline plasma cell of >20% than in those with a lower plasma cell percentage (<10% and 10-20%). However, in patients treated with D-VCd, haematologic CR was higher regardless of the baseline plasma cell percentage (Table 6). Little effect of baseline plasma cell percentage was observed on organ response rates.

Discussion
The primary results from the ANDROMEDA study demonstrated that in patients with newly diagnosed AL amyloidosis, the addition of daratumumab to VCd was superior to VCd alone, with higher rates of haematologic CR at a median follow-up of 11.4 months, cardiac and renal responses at 6 months, and improved major organ deterioration-PFS [10]. This post hoc analysis of ANDROMEDA evaluated clinical outcomes in patients with 4 common cytogenetic abnormalities seen in patients with AL amyloidosis: t (11;14), del17p13, del13q14, and amp1q21.
This large analysis from a prospective clinical trial validates the results from previous studies demonstrating that t(11;14) was the predominant cytogenetic abnormality in this patient population [2,17,18]. Outcomes from this  6 months (B). a For the subgroup analyses, the OR and 95% CI were from unstratified Mantel-Haenszel estimate of the common OR. For the ITT population, Mantel-Haenszel estimate of the common OR for stratified tables was used. The stratification factors from the interactive web response system were: cardiac staging (I, II, IIIa), countries that typically offer or not offer transplant for patients with AL amyloidosis (List A, List B), and baseline renal function (CrCl !60 mL/min or CrCl <60 mL/min). b P-value from Fisher's exact test. c P-value from the Cochran-Mantel-Haenszel Chi-Squared test. d At least two abnormal genes among: del17p13, t (11;14), del13q14, and amp1q21. Twelve-month landmark data cut (CCO November 2020). CI: confidence interval; CR: complete response; D-VCd: daratumumab, bortezomib, cyclophosphamide, and dexamethasone; ITT: intent to treat; NE: not evaluable; OR: odds ratio; VCd: bortezomib, cyclophosphamide, and dexamethasone. Figure 3. Rate at 6 months of (A) cardiac response (cardiac response-evaluable) and (B) renal response (renal response-evaluable) with D-VCd and VCd across cytogenetic subgroups. a For the subgroup analyses, the OR and 95% CI were from unstratified Mantel-Haenszel estimate of the common OR. For the ITT population, the Mantel-Haenszel estimate of the common OR for stratified tables was used. The stratification factors from the interactive web response system were: cardiac staging (I, II, IIIa), countries that typically offer or not offer transplant for patients with AL amyloidosis (List A, List B), and baseline renal function (CrCl !60 mL/min or CrCl <60 mL/min). b P-value from Fisher's exact test. c P-value from the Cochran-Mantel-Haenszel Chi-Squared test. d At least two abnormal genes among: del17p13, t (11;14), del13q14, and amp1q21. Twelve-month landmark data cut (CCO November 2020). CI: confidence interval; D-VCd: daratumumab, bortezomib, cyclophosphamide, and dexamethasone; ITT: intent to treat; OR: odds ratio; NE: not evaluable; VCd: bortezomib, cyclophosphamide, and dexamethasone. analysis in patients with cytogenetic abnormalities were generally consistent with the ITT population and the primary results from ANDROMEDA [10]. Rates of haematologic CR at a median follow-up of 20.3 months and at 6 months, and cardiac and renal responses at 6 months were higher in patients treated with D-VCd vs VCd across all cytogenetic subgroups except for the cardiac response in the del17p13 group, and these outcomes were comparable to the overall ITT population. In addition, more patients achieved a deep response with D-VCd than VCd for the ITT population, and the t(11;14) subgroup and point estimates of HR for major organ deterioration-PFS and -EFS also favoured D-VCd over VCd across all cytogenetic subgroups and in the ITT population.
The results from this analysis indicate that in patients with any of the 4 cytogenetic abnormalities, treatment with VCd was associated with suboptimal haematologic CR (0-14.3%) while the benefit seen with D-VCd vs VCd was unaffected. These results validate previous studies where poorer and less deep responses have been reported in patients with AL amyloidosis and t(11;14) who were treated with bortezomib-based regimens [8,9,19]. Given the poor outcomes of patients with newly diagnosed AL amyloidosis and 1 of the 4 cytogenetic aberrations who were treated with VCd alternative treatment options should be considered. Aside from the use of D-VCd, another potential treatment option for these at-risk patients may be venetoclax, which currently shows promise for the treatment of patients with relapsed/refractory AL amyloidosis and t(11;14) [20].
Patients with a diagnosis of MM were excluded from this study, but it is generally accepted that the presence of certain cytogenetic abnormalities, 2 or more in particular, can be predictive of poorer outcomes in these patients, and cytogenetic testing can be used to help guide treatment choices [21,22].
This study does have some limitations. The subgroup analysis was not prespecified and the sample sizes are small. In addition, FISH/karyotyping was performed locally as opposed to centrally. Not unexpectedly, the del17p13 subgroup was particularly small, consisting of only 18 patients (9 in each treatment group). In addition, the proportion of patients with del13q14 was lower (19%) than is generally reported in the literature (approximately one-third) [2]. This may have been the result of the exclusion of patients with MM from this study. Most case studies of amyloidosis also include patients with MM, a population in whom the prevalence of del13q14 would be expected to be higher ($40-50%) [23,24]. In addition, the use of karyotyping as well as FISH to identify the abnormalities may have impacted the sensitivity of the testing. Another limitation of this analysis is the short follow-up.
In conclusion, the results of this post hoc analysis of the ANDROMEDA study are consistent with the primary analysis. Rates of overall haematologic CR and cardiac and renal responses at 6 months were numerically higher with D-VCd than VCd, and major organ deterioration-PFS and -EFS favoured D-VCd over VCd across all cytogenetic abnormalities evaluated. Rates of haematologic CR in patients treated with D-VCd were unaffected by the Figure 5. Subgroup analysis of major organ deterioration-EFS. HR and 95% CI for each cytogenetic subgroup are from an unstratified Cox proportional hazards model with treatment as the sole explanatory variable. HR and 95% CI for the ITT population are from a Cox proportional hazards model with treatment as the sole explanatory variable and stratified with cardiac stage (Stage I, II, and IIIa), countries that typically offer or not offer transplant for patients with AL amyloidosis, and renal function (CrCl !60 mL/min or CrCl <60 mL/min) as randomised. Primary data cut (CCO Feb 2020). CI: confidence interval; D-VCd: daratumumab, bortezomib, cyclophosphamide, and dexamethasone; EFS, event-free survival; EVT: event; HR: hazard ratio; ITT: intent to treat; NE: not evaluable; VCd: bortezomib, cyclophosphamide, and dexamethasone. Twelve-month landmark data cut (CCO November 2020). D-VCd: daratumumab, bortezomib, cyclophosphamide, and dexamethasone; ITT: intent to treat; VCd; bortezomib, cyclophosphamide, and dexamethasone. Table 4. Subsequent lines of therapy by cytogenetic abnormality and treatment. Not including alkylating agents. Twelve-month landmark data cut (CCO November 2020). D-VCd: daratumumab, bortezomib, cyclophosphamide, and dexamethasone; ITT: intent to treat; LOT: line of therapy; VCd: bortezomib, cyclophosphamide, and dexamethasone. presence versus absence of t(11;14) but were lower in patients with versus without t(11;14) who were treated with VCd. These findings add to the growing body of evidence supporting the use of cytogenetic testing when targeting treatment choices in patients with AL amyloidosis. Additional large, randomised studies are needed to further explore longer-term outcomes in patients with AL amyloidosis in the context of cytogenetic abnormalities.

Ethical statement
The study protocol was approved by each study site's local independent ethics committee or Institutional Review Board, and the study was conducted in accordance with the principles of the Declaration of Helsinki and the International Conference on Harmonisation Good Clinical Practice guidelines. All patients provided written informed consent. A. Dispenzieri has received research funding from Alnylam, Celgene, Intellia, Janssen, Pfizer, and Takeda.
D. Bhutani has received clinical trial funding from Sanofi pharmaceuticals and served as a consultant for Sanofi Genzyme.
M. Gertz has received honoraria from Spectrum, Janssen, Celgene, Alnylam, and Ionis, and research funding from Spectrum.
A. Wechalekar has received honoraria from Celgene, Janssen, and Takeda, served on advisory boards for Caelum and Janssen, and received travel funding from Takeda.
G. Palladini has served on the advisory boards of Alexion, Argobio, Janssen, and Protego, received honoraria from Alexion, Argobio, Janssen, Pfizer, Protego, Prothena, Sebia, Siemens, and The Binding Site, and received research funding from Gate Bioscience and The Binding Site.
A. Jaccard has served in consulting or advisory roles for Janssen and received honoraria and research and travel funding from Janssen.
E. Kastritis has served in a consulting role for Amgen, Genesis Pharma, Janssen, Pfizer, GSK, Prothena and Takeda, received honoraria from Amgen, Genesis Pharma, Janssen, Pfizer, GSK and Takeda, and received research funding from Amgen and Janssen.
S. Sch€ onland received honoraria and travel support or research funding by Janssen, Prothena, and Takeda.
C. la Porte was employed by Janssen at the time of this research. H. Pei and N.P. Tran are employed by Janssen and hold company stock.
G. Merlini has nothing to disclose. Figure 6. Proportion of patients in the ITT and t (11;14) populations with a deep response. a a Deep response defined as patients achieving either dFLC <10 mg/L or iFLC 20 mg/L. Twelve-month landmark data cut (CCO November 2020). D-VCd: daratumumab, bortezomib, cyclophosphamide, and dexamethasone; ITT: intent to treat; VCd: bortezomib, cyclophosphamide, and dexamethasone.