Prognostic Implication of Bone Marrow Abnormality of Appendicular Skeletons Detected By Low-Dose Multidetector Computed Tomography in Patients with Myelodysplastic Syndrome

Background

A number of prognostic systems have been developed and validated in patients with myelodysplastic syndrome (MDS). However, the use of the current scoring systems including international prognostic scoring system (IPSS) in determining the prognosis among MDS patient has various conflicting views pertaining to the inclusion of patients with different outcomes. Abnormal red marrow distribution in appendicular skeletons (AS) based on magnetic resonance imaging (MRI) has been described in the literature; however, these findings have not been confirmed. We examined the prognostic implication of medullary abnormalities of AS detected by low-dose whole-body multidetector computed tomography (MDCT) in patients with MDS.

Patients and methods

147 consecutive patients with MDS diagnosed between January 2009 and May 2017 at Kameda Medical Center were recruited. Diagnosis and classification of MDS were made per 2008 WHO criteria. MDCT studies were performed at the diagnosis of MDS and before the start of any chemotherapy. MDCT images were acquired as previously reported. Medullary pattern of AS were categorized into three groups: (1) Fatty marrow pattern; (2) Focal/scattered marrow pattern; (3) diffuse marrow pattern. To examine if the value of MDCT findings on appendicular marrow has incremental predictive value when the conventional risk factors were controlled, we constructed a receiver operating characteristics (ROC) curve for two logistic regression models. Model 1 was constructed with only conventional risk factors while Model 2 was constructed with conventional risk factors plus MDCT findings.

Results

Of the 147 patients with MDS, there were 47 (32.0%) with fatty, 51 (34.7%) with focal/scattered and 49 (33.3%) with diffuse marrow patterns. Fatty pattern was seen more among non-RAEB patients (76.5%), focal/scattered pattern was about equal while diffuse pattern was more in RAEB patients with 65.3% (P<0.001). The development of AML differed significantly between groups and the highest occurred with diffuse pattern, 30 (61.2%), (P<0.001). Among those with AML, death occurred among 28 (93.3%) . The median LFS was not reached (NR) for fatty pattern, but was 29.2 and 11.1 months for the focal/scattered and diffuse marrow patterns respectively (P<0.001). Similarly, the median OS was NR for fatty pattern but was 40.9 and 14.6 months for the focal/scattered and diffuse marrow patterns respectively (P<0.001). Among the 82 low-risk patients, median survivals of fatty, focal/scattered, and diffuse pattern differed significantly and were NR, 35.2 and 18.2 months respectively for LFS (P<0.001), and NR, 50.9 and 26.3 months respectively for OS (P<0.001). Similarly, for the 65 patients with high-risk MDS, median time were NR, 11.2 and 8.3 months for LFS (P<0.001), and NR, 19.1 and 11.4 months for OS (P<0.001), respectively for fatty, focal/scattered, and diffuse patterns. At multivariate analysis, both focal/scattered and diffuse patterns of medullary marrow of AS remained independently associated with shorter LFS and OS. ROC curve was developed based on the conventional risk model as well as the conventional risk model plus MDCT findings for LFS and OS. Although AUC comparisons showed statistical significance in neither LFS (P=0.108) nor OS (P=0.138), net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were significant for both LFS (NRI 0.43 [95% CI;0.12-0.74], P=0.007 and IDI 0.06 [95% CI; 0.02-0.10], P=0.002) and OS (NRI 0.52 [95% CI; 0.21-0.83], P<0.001 and IDI 0.07 [95% CI; 0.03-0.11], P<0.001).

Conclusion

Our findings showed a novel method for prognostication of MDS using the medullary findings of AS using MDCT. The prognostic impact of medullary abnormality of AS detected by MDCT has not been described previously. Our findings indicate that the MDCT of appendicular bone marrow is predictive of the prognosis in patients with MDS and has a potential for improving the risk stratification of MDS in conjunction with preexisting scoring systems.

View largeDownload slide

View largeDownload slide

Close modal