Myeloma bone disease imaging on a 1st-generation clinical photon-counting detector CT vs. 2nd-generation dual-source dual-energy CT

Multiple myeloma is a malignant hematologic disease of the mature B-cells primarily affecting the bone marrow. However, early in the disease course, medullary tumor cell expansion paired by complex paraneoplastic pathophysiologic mechanisms affecting bone metabolism aggresses also the bone (myeloma bone disease) leading to bone destruction in a diffuse or focal manner, mostly combined [1]. Affection of the skeleton has two major implications [2]. One is the tumor staging according to the Salmon and Durie classification from 1975 (bony lysis as a surrogate marker for marrow infiltration; staging depends on the number of lesions) which has immediate therapeutic implications whereas the other one is related to the risk of fracture and related neurologic complications [3]. For this purpose, an accurate skeletal evaluation is mandatory. Classically, this task was managed by plane radiographs, but in modern times CT has advanced to the method of first choice due to its superior spatial resolution and absence of image superimposition [4‐6]. Nevertheless, even CT has its limitations which are in part methodically inherent (inferior bone marrow sensitivity, e.g., over MRI) and in part derived from protocol constrains (low dose, high pitch, low spatial resolution, etc.) [7, 8]. However, a more accurate and earlier detection of bone abnormalities would be desirable for decision making. As whole-body CT-surveillance implies a large field-of-view, in-plane resolution is thereby diminished whereas the use of comb filters to achieve high-resolution imaging implies decreased radiation dose efficiency [9].

At this point, the advent of photon-counting CT-technology might represent a promising approach to overcome these limitations. The photon-counting detector measures each individual x-ray that passes through the patient’s body eliminating electronic noise and is, thus, presumed to deliver comparable image quality with that of previous generations of dual-source dual-energy integrating detector CT (DS-EID-CT) by significantly reduced radiation exposure or to increase image quality including spatial resolution by comparable dose exposure [10‐14]. It has smaller detector elements which improve visualization of fine details (e.g., in the cancellous bone) reducing simultaneously the image noise. Other contributors to the superior image quality of DS-PCD-CT include improvements in noise reduction algorithms and iterative reconstruction techniques [15, 16]. Small detector element size may translate into improved visualization of fine detail and image noise reduction through anti-alising, particularly for high-spatial resolution images as with our protocol [17, 18].

This study, to the best of our knowledge, is the first to address the potential superiority of PCDs over energy-integrating detectors (EID) for increased spatial resolution of bone microstructure including disease-related bony defects in the clinical workup of multiple myeloma patients.

Between October 2021 and February 2022, 50 myeloma patients (34 male, 16 females, mean age: 67.7 ± 10.9 years) were included. Hematologic diagnosis based on M gradient was IgG (n = 24), IgA (n = 9), light chain MM (n = 14), and non-secretory MM (n = 3) (Table 2). The median time interval between clinically indicated DS-EID-CT and DS-PCD-CT was 12 months (range: 4–81 months). The radiation exposure associated with DS-EID-CT resulted in a mean dose-length product (DLP) of 1344.3 ± 204.6 mGy*cm and mean volume mean volume CT dose index (CTDIvol) of 10.1 ± 1.9 mGy. Radiation exposure related to DS-PCD-CT had a mean DLP of 1107.4 ± 247.6 mGy*cm and CTDIvol of 8.2 ± 1.8 mGy. Elevated laboratory values of serum IgG, IgA, serum or urine light-chain λ, or light-chain κ alone or in combination at date of examination were identified in 17 patients (34%) prior to DS-EID-CT and in 16 patients (32%) prior to DS-PCD-CT (p = 1.0). Urine M-protein was increased in 9 patients (18%) before DS-EID-CT and in 6 patients before DS-PCD-CT (p = 0.453). All patients with active disease experienced minimal changes in the serum (p = 1.0) and urine levels of M-gradient (p = 0.453) at interval. Most patients were in complete remission or had no disease activity at the time of DS-EID-CT (n = 32) and DS-PCD-CT (n = 33), and non-parametric analyses revealed no significant difference in disease status (complete response, partial response, and disease progression) between the two time points (p = 0.308). All patients with lytic bone lesions received monthly bisphosphonates. At the time of DS-EID-CT, 21 patients (42%) and during DS-PCD-CT, 19 patients (38%) received immune-based therapy. Three patients (6%) were respectively undergoing intensive cytotoxic chemotherapy at the time of DS-EID-CT and DS-PCD-CT. There was no significant difference in the distribution of the different types of treatment (p = 0.608) and CT imaging features (p = 1.0) between the two study time points. Clinical parameters on both examination days are summarized in Table 3.

In this study, an ultra-high-resolution scan mode on a DS-PCD-CT scanner with a detector pixel size of 0.2 mm at the isocenter was compared with a DS-EID-CT with a detector pixel size of 0.2 × 0.2 mm² for whole-body skeletal imaging in patients with multiple myeloma focusing on the trabecular microstructure of the cancellous bone and on the disease-related lytic bone lesions. The scan parameters were set as far as possible at similar levels for both scanners in order to achieve similar radiation doses which finally did not differ significantly from each other although they proved slightly lower (10–15%) for the DS-PCD-CT. As on DS-EID-CT only the sagittal and coronal MPRs were reconstructed using a sharp kernel, we decided to focus our image analysis for both scanners on these reformates due to the better comparability of these image kernels. Most other adjustable parameters were also similar.

The results of our study show significant qualitative and quantitative image quality improvement on the DS-PCD-CT compared to the DS-EID-CT. Hence, the delineation of bony trabeculae as well as the sharpness of their edges and the transitions between the cancellous bone and focal lytic bone lesions was more easily and reliably accomplished on DS-PCD-CT. Notably, this trend stayed unimpaired even in bone lesions smaller than 5 mm. Despite the use of a sharp kernel and submillimeter single slice collimation, depiction of densely packed trabeculae in the femoral heads exhibited a blurred, unsharp appearance on the DS-EID-CT. Knowingly, the trabecular density varies significantly between subjects being dependent on many factors like age, gender, bone metabolism, medication, physical activity, etc. Therefore, only an intra-individual comparison is allowed in such cases which we have accomplished in this study showing an interreader agreement of between 94.2 and 98.8%.

For more objective data analysis, we additionally compared the delineation (sharpness) of the contours of lytic bone lesions by the maximal slope within CT value plots and found a significantly steeper slope with DS-PCD-CT as with DS-EID-CT. Here again, quantitation of the number of serrations (trabecular structures) was significantly higher for DS-PCD-CT compared to DS-EID-CT and these results were proven once more independent on the lesion size. This latter aspect may lead in the clinical routine while comparing CT-images from different scanners to the false visual impression of detecting more bone lesions on the DS-PCD-CT because they become more conspicuous. A side-by-side analysis of submillimeter axial sizes was used for confirmation and or exclusion of new osteolysis in our study.

Improvement of spatial resolution has been always in the focus of radiological imaging. Different approaches have been proposed and tested in the clinical routine; however, the main limitations with the conventional DS-EID-CT are the radiation dose and the size of the detector elements. The photon-counting detectors have opened new possibilities for improved spatial resolution without loss of dose efficiency. Conducting a study on a DS-PCD-CT using two high-resolution scan modes (sharp and UHR), Leng et al reported a 69% and 87% improvement in in-plane resolution over an EID-scanner for a detector pixel size of 0.25 mm at the isocenter [12]. The use of filters consisting of many septa decreases the size of detector elements with EIDs, but at the same time it increases dose as the blocked photons have already contributed to patient dose, but not to the reconstructed image. On the contrary, no septa are present in DS-PCD-CT and therefore reduced efficiency is not expected. The superiority of the DS-PCD-CT scanner for decreasing image noise could be already demonstrated in a cadaveric study by Zhou et al, focusing on temporal bone imaging [22]. These authors indicated the potential of a 64% reduction in dose using a DS-PCD-CT UHR mode for clinical bone imaging. Besides first attempts to improve spatial resolution with aid of PCD-technology for smaller structures (small FOV), Rajendran et al reported their initial experience with a DS-PCD-CT using a full FOV in UHR mode (0.2 mm resolution) [23]. Their study reported a 37% lower radiation dose and 46% lower image noise compared with EID-CT. The authors thus demonstrated that larger body parts can be imaged in the UHR mode. Similarly, but in a phantom study, Yu et al investigated and confirmed the clinical feasibility of a whole-body DS-PCD-CT-scanner [24]. Using the same quantification software, Bette et al, in an animal study, demonstrated the subjective and objective image quality superiority of DS-PCD-CT over DS-EID-CT for delineation of tiniest bone details [13]. Thus, PCD improves image quality and also spatial resolution because of the more efficient conversion of x-rays into light and by minimizing electron noise [25]. Also improvements with the iterative reconstruction techniques contribute to the observed improvements in image quality [15]. In line with these previous reports, our study shows that a significant improvement in spatial resolution is possible with the new DS-PCD-CT by comparable or lower radiation dose as with DS-EID-CT even for whole-body studies with full FOV.

Our study has some limitations. First, the number of patients was too small in order to allow for a more sophisticated subgroup analysis of patients with different degrees of disease activity affecting the contrast between the infiltrated bone marrow in the background and the focal osteolysis. Second, we did not have additional information on the underlying bone density of our patients which would be expected to also impact the contrast between bone marrow and trabeculae of the cancellous bone as well as their thicknesses. Third, slight differences with respect to the used kernels as well as the matrix could have had a certain impact on image comparability, but at least the former is no more available for the DS-PCD-CT. We used MPRs to mitigate the differences in the kernels, but in either the x- or y-direction, the image impression is still dominated by the kernel.

To the best of our knowledge, this study is the first to confirm the superiority of DS-PCD-CT over DS-EID-CT for assessment of bone microstructure in humans, in particular in the clinical setting of myeloma bone disease.

In conclusion, DS-PCD-CT significantly improves spatial resolution of bony microstructure and disease-related (lytic bone lesions) compared to 2nd-generation DS-EID-CT.