Current Staging Systems Are Inadequate Predictors of Treatment Failure after Autologous Hematopoietic Stem Cell Transplants (AuHCT) in Multiple Myeloma (MM).

The new international (ISS) staging system is easier to compute and a better predictor of survival after diagnosis (dx) of MM than the Durie Salmon staging system (DSS). It is not known if either system predicts treatment failure following upfront AuHCT. We compared the ISS vs DSS at dx for predicting progression free survival (PFS) after AuHCT. The outcomes of 729 recipients of AuHCT done ≤12 months of dx reported to the CIBMTR between 1995–2002 were analyzed. Treatment failure (progression or death) after AuHCT was modeled using Cox proportional hazards regression. Risk factors considered were age, gender, Karnofsky score, immunochemical subtype, number of lines of chemotherapy, chemotherapy sensitivity, disease status pre transplant, serum creatinine, time from dx to transplant and year of transplant. Cox models were then fit with DSS and ISS stages in addition to covariates identified as significant. The percentage of explained variability (R²) was calculated using Brier scores. The Brier score estimates the predictive performance of prognostic schemes for survival data. A smaller Brier score and a larger R² indicate better predictive performance. 61% of patients were male; 62% had performance scores ≥90 and 82% had chemosensitive disease. Single (91%) or tandem (9%) AuHCT was performed at a median of seven mo after dx; 73% used melphalan-based regimens. With a median follow up of 56 mo after AuHCT, univariate probabilities of treatment related mortality (TRM), relapse, PFS and overall survival at 5 years were 7 (95% confidence interval [CI], 5–9)%, 68 (64–72)%, 25 (21–29)% and 52 (47–56)%, respectively. In multivariate analysis, significant risk factors for treatment failure were resistance to chemotherapy pre transplant, no remission pre transplant and longer delay from dx to transplant. Most patients were in DSS stage III (62%). Patients were more evenly distributed across the 3 ISS stages; only 36% were concordant between systems (Table 1). Cohen’s kappa (a measure of agreement between scores) indicated little concordance between DSS and ISS (K=0.09 [95% CI: 0.04–0.13]). Both systems suggested patients well for PFS and OS. Median PFS times for stages I, II and III by DSS and ISS systems were 58, 31, 24 mo and 33, 27 and 23 mo, respectively. Corresponding median survival times for stages I, II, III by DSS and ISS were 82, 68, 50 and 64, 68, 45 mo, respectively. Compared to the ISS, DSS demonstrated a lower BS and higher R² (Table 2) consistent with a marginally superior predictive ability for 1, 3 and 5 year PFS (Figure). However, neither system is strongly predictive of outcome after AuHCT since the R² (measure of explained variability) is low (<5%) for both systems - indicating the need for newer schemes that incorporate other predictors, e.g. cytogenetics.

Table 1

.	ISS, N(%) .
DS, N(%)	I	II	III
I	26 (4)	20 (3)	4 (1)
II	92 (13)	104 (14)	34 (5)
III	134 (18)	183 (25)	132 (18)

.	ISS, N(%) .
DS, N(%)	I	II	III
I	26 (4)	20 (3)	4 (1)
II	92 (13)	104 (14)	34 (5)
III	134 (18)	183 (25)	132 (18)

View Large

Table 2

.	BS .		R2,% .
Time .	DSS .	ISS .	DSS .	ISS .
1 year	0.1855	0.1860	4.1	3.9
3 years	0.2323	0.2343	4.2	3.4
5 years	0.1818	0.1846	2.7	1.2

.	BS .		R2,% .
Time .	DSS .	ISS .	DSS .	ISS .
1 year	0.1855	0.1860	4.1	3.9
3 years	0.2323	0.2343	4.2	3.4
5 years	0.1818	0.1846	2.7	1.2

View Large

View largeDownload slide

Figure

View largeDownload slide

Figure

Close modal