Cytogenetic prognostic index for survival in multiple myeloma

Patients with multiple myeloma (MM) can be defined as having low, intermediate and high-risk cytogenetics. Current definitions are based on a small number of common cytogenetic abnormalities, considered by some to be oversimplified. Additionally, patients with deletion of chromosome 17p (del[17p]) and translocation (4;14) (t[4;14]) have a heterogenous survival profile which indicates that additional factors must influence outcome.¹

In recent years, there has been increasing interest in personalized therapy for MM based on disease characteristics. Several genetic abnormalities are accepted to impact survival outcomes in MM, in relation to overall survival (OS) and progression-free survival (PFS), and these are summarized in Table 1.¹

Table 1. Commonly accepted chromosomal changes in MM and their impact on outcome¹

Abnormality	Approximate % of patients with NDMM affected	Impact on outcome (PFS and OS)
del(17p)	8%	Negative
t(4;14)	15%	Negative
del(1p32)	8%	Negative, particularly when t(4;14) also present
Gain 1q21	33%	Negative
Trisomy 21	-	Negative
Trisomy 3	-	Positive
Trisomy 5	-	Positive

In clinical trials, specific cytogenetic subgroups of patients have exhibited a survival benefit with different drugs and regimens, indicating personalizing treatment based on individual risk may be a viable strategy. A subgroup analysis of the BELLINI study found that patients with relapsed/refectory MM with t(11;14) benefitted from venetoclax + bortezomib + dexamethasone (VenBd) treatment, without associated treatment toxicity, compared to other patient populations.² In the MM-003 study, pomalidomide + low-dose dexamethasone was found to be of particular benefit for patients with del(17p), giving outcomes comparable to standard-risk patients.³ In contrast, other studies have not found a clinical benefit of certain regimens in different molecular subgroups. One example is the Myeloma XI trial which found no difference in response to treatment in a patient subgroup analysis.⁴

Reaching a consensus on the definition of cytogenetic risk could assist in personalizing MM therapy based on individual patient characteristics. In a paper recently published in the Journal of Clinical Oncology, Aurore Perrot, Centre Hospitalier R´egional Universitaire Nancy, Nancy, FR, and colleagues, simultaneously investigated the impact of seven common genetic abnormalities, in a cohort of patients with newly diagnosed multiple myeloma (NDMM). Their aim was to develop and validate a cytogenetic prognostic index (PI), to ease the stratification of patients into risk groups.¹

Study design and patient characteristics¹

This study evaluated the prognostic impact of seven genetic abnormalities; del(17p), t(4;14), del(1p32), 1q21 gain and trisomies 3, 5 and 21, using fluorescence in situ hybridization (FISH) and single nucleotide polymorphism (SNP).

In total, data from 1,635 patients was obtained from four trials by the Intergroupe Francophone du Myélome (IFM) including IFM 99-02, IFM 99-06, IFM 2005-01 and IFM 2009. These patients were then split into a training set, an internal validation set, and two external validation sets (Table 2). The external validation sets used patients for whom international stating system (ISS) disease stage information was available and who had bone marrow plasma cells available in the IFM myeloma biobank. The second external validation set used patient data from the more recent IFM 2009 study.

Table 2. Datasets used for cytogenetic analysis

		Validation data sets
	Training set	Internal	External 1	External 2
Purpose of model	Model development	Testing the model	Testing the model in patients with ISS disease stage information available. Evaluation of prognostic capability of model in patients treated with newer, more effective, treatment regimens
N	720	234	359	322
Presence of the following cytogenetic abnormalities:
t(4;14)	19.5%	17.1%	13.1%	9.9%
del(17p)	13.9%	9.8%	13.1%	7.8%
Trisomy 5	37.2%	41.9%	37%	43.2%
Trisomy 21	23.6%	22.6%	26.5%	25.5%
1q gain	40%	36.8%	36.8%	32.3%
del(1p32)	12.4%	12.4%	8.1%	8.1%
Median follow-up (years)	8.2	6	7.3	4.9
Estimated 5-year survival probability	58%	62%	54%	80%
Median number of abnormalities (interquartile range)	2 (1–3)	1 (1`–2)	1 (1–2)	1 (1–2)
No genetic abnormality	20%	20%	25%	25%

In the training set, 74% of patients received intensive treatment with chemotherapy, high-dose melphalan (HDM) and autologous stem cell transplant (ASCT), with 26% of patients being over the age of 65 and receiving non-intensive treatment. Treatments were similar in the first external validation set. However, patients in the second validation set received the newer triplet of bortezomib, lenalidomide and, dexamethasone (VRd), with or without HDM + ASCT, allowing the study investigators to asses the PI validity on a new treatment population.

Results¹

Six abnormalities were statistically relevant based on multivariate cox proportional hazard regression modelling (Table 3), and these were incorporated into the PI calculation:

0.4 x t(4;14) + 1.2 x del(17p) – 0.3 x trisomy 5 + 0.3 x trisomy 21 + 0.5 x 1q gain + 0.8 x del(1p32)

Table 3. Multivariate cox proportional hazard regression model for myeloma-specific survival stratified by treatment in the training set in the final prognostic model

Abnormality	β (95% CI)	Hazard Ratio (HR, 95% CI)	p
t(4;14)	0.40 (0.15–0.67)	1.50 (1.16–1.95)	0.002
del(17p)	1.17 (0.89–1.44)	3.21 (2.44–4.22)	< 0.001
Trisomy 3	-	-	-
Trisomy 5	-0.35 (-0.59– -0.11)	0.71 (0.56–0.9)	0.005
Trisomy 21	0.34 (0.09–0.59)	1.41 (1.09–1.81)	0.008
1q gain	0.5 (0.3–0.7)	1.64 (1.34–2.01)	< 0.001
del(1p32)	0.8 (0.54–1.05)	2.21 (1.71–2.87)	< 0.001

Using this score, three risk groups were identified, each with a different survival estimate (Table 4).

Table 4. Risk groups by PI identified in training set

Risk group	PI score	Estimated 5-year survival
Low	≤0	>75%
Intermediate	0–1	50–75%
High	> 1	<50%

Tables 5 shows the associated myeloma-specific survival for each identified subgroup in each dataset.

Table 5. Myeloma-specific survival by cytogenetic PI

	Low-risk	Intermediate-risk	High-risk
Training set (n= 647)	HR= 1.00	HR= 2.51 95% CI, 1.83–3.44 p< 0.001	HR= 6.41 95% CI, 4.57–8.99 p< 0.001
Internal validation (n= 234)	HR= 1.00	HR= 3.22 95% CI, 1.79–5.78 p< 0.001	HR= 11.44 95% CI, 6–21.82 p< 0.001
External validation 1 (n= 359)	HR= 1.00	HR= 5.27 95% CI, 3.38–8.21 p< 0.001	HR= 15.22 95% CI, 9.34–24.82 p< 0.001
External validation 2 (n= 322)	HR= 1.00	HR= 5.23 95% CI, 2.63–10.40 p< 0.001	HR= 8.13 95% CI, 3.72–17.74 p< 0.001

In the training set, a PI of 0 was only found in patients with no abnormalities. In all data sets, a high PI was associated with poor survival, particularly in intermediate and high-risk group compared to low-risk. Hazard ratios for death for those in the high-risk group were six to 15 times higher than those in the low-risk group.

The results above were confirmed in both external validation data sets, after ISS disease stages were included. Additionally, the cytogenetic PI was able to predict patients who survived, in all data sets.

Conclusion¹

The authors of this study have developed and validated a cytogenetic PI which improves the classification of patients with NDMM, compared to current models.

Advantages of this new cytogenetic PI are that it is a weighted score, accounting for multiple cytogenetic abnormalities, and includes a positive prognostic factor of trisomy 5. Additionally, the model was validated in two external datasets in which patients were treated with newer induction and consolidation therapies, such as VRd, as well as effective second-line treatment. Significantly, the model remained prognostic and discriminative when tested in these external validation cohorts, despite patients in these groups having a higher probability of survival due to the use of newer treatment strategies. However, since this PI was developed based on patients enrolled in clinical trials, there is a limitation for translating this into daily clinical practice.

In conclusion, this new PI may facilitate more effective clinical trials, allowing for risk-adapted treatments more personally tailored to the patient to reach the clinic faster.