The present study provides clinical and imaging findings of MS based upon a large patient sample in a multicenter design.
Making the diagnosis of MS can be a challenge, especially in patients with extramedullary MS and normal appearing bone marrow as a first manifestation of AML [
6]. These patients represent, however, less than 1% of all AML cases [
6]. MS is associated with an overall poor survival with reported median survival of 12.8 to 15.9 months [
6,
8,
12]. In short, sole presence of MS indicates poor outcome irrespective of the clinical setting.
Clinical findings
Principally, MS can be classified into 4 groups according to the manifestation. First is the manifestation of MS with concurrent acute myeloid leukaemia. In such cases the diagnosis of MS might be easy to make [
5,
12,
13]. Secondly, extramedullary relapse of AML, including in the setting after bone marrow transplantation. Thirdly, blast phase/transformation of a myeloproliferative neoplasm or chronic myelomonocytic leukaemia [
5,
12,
13]. Lastly, isolated MS, which occurs in association with a normal bone marrow biopsy and blood analysis, and in the absence of any history of myeloid neoplasia [
6]. These cases of MS might be most challenging in diagnosis and needs histopathological evaluation of the MS suspicious lesion [
6,
12,
13].
The identified frequencies in the present study are well comparable with the literature. Most common the extramedullary relapse setting of AML with up to 60%, followed by simultaneous manifestation in 30% and less frequently, as an extramedullary manifestation of AML without bone morrow involvement in only few cases.
The diagnosis of MS can be challenging, and relatively high misdiagnosis rates have been reported varied from 25 to 47%, which especially concern de novo manifestations without bone marrow involvement [
12‐
14]. Possible misdiagnoses comprised Hodgkin-Lymphomas, MALT lymphoma, or Ewing’s sarcoma, which can show similar histopathologic characteristics [
12].
Clinically, infectious diseases such as abscesses or hematomas should be considered as a possible differential diagnosis because these occur very frequently in leukemic patients with immunosuppression and thrombocytopenia, either due to chemotherapy or due to malignancy itself [
8].
Regarding gender, a slight male predominance was identified in the literature [
6,
13]. However, in the present sample there was no gender predominance.
Clinical presentation of MS largely depends on the affected site. Correspondingly, MS can present with various symptoms, such as tumor mass effect or local organ dysfunction [
2]. However, according to the literature, about half of patients with MS were asymptomatic and, therefore, they were identified by imaging [
8]. Notably, in the present study, most cases of MS (24.5%) were detected incidentally by imaging studies, which is caused by increasing usage of cross-sectional imaging in hematological patients. Therefore, the radiologist and oncologist need to be aware of MS.
Regarding localizations, MS affects most frequently cutis, subcutis and lymph nodes [
5,
9,
13]. However, the frequencies of different localizations varied in several studies. In fact, Kaur et al. reported a skin involvement in up to 69.5% of patients in a case series of 22 patients [
5]. Contrary, Pileri et al. reported a skin manifestation in only 28.2% of patients based upon 74 patients, which was yet the most common localization [
13]. Kawamoto et al. reported clinicopathological findings of 131 patients with MS and identified that lymph nodes were the most frequent manifestation (55%) [
3]. Recently, a frequent involvement of the visceral soft tissue (29.9%) as the most common localization was described [
9]. MS can principally manifest in every organ resulting in rare organ manifestations including pulmonary, kidney, vaginal and uterine [
8]. In the present study, most commonly cutis involvement was found, followed by bone and lymphatic tissue.
A relatively high amount of breast manifestations was identified in our patient sample. Contrary, in previous patient samples no breast manifestation was reported. This might be caused by a university hospital included in the study with a large breast center and therefore possible selection bias.
The affection of the central nervous system is rare with a reported frequency of 1.5%, which can have a crucial impact on the clinical course of patients due to early neurological deficits [
3]. We identified a higher rate, namely 9.3% of all acquired cases. Presumably, this might be caused by an increasing use of cross-sectional imaging in oncologic patients, which leads to more incidentally detected lesions. This might explain differences to older case series with lower reported frequency [
15].
Imaging findings
Previously, only few reports with relatively small numbers of patients/lesions reported imaging findings in MS [
9‐
11,
16,
17]. So far, Shinagare et al. described MRI features of 25 patients with 41 different MS localizations [
16]. The authors reported that the lesions had a mean size of 5.6 cm (range 1–20 cm), which is slightly higher than in our observation. Presumably, the results differ slightly due to the fact that in the mentioned study only lesions with MRI were included, whereas in the present study mostly CT was used. A MS lesion investigated by MRI might be a clinical symptomatic lesion with consecutive a larger size, whereas CT more commonly detects incidental lesions, which might be smaller in size.
CT findings of MS were reported to be variable, depending on the site of involvement [
8]. Most commonly, on CT images, MS lesions were reported to be isodense to adjacent muscle tissue [
9]. In contrast, cerebral manifestations were reported to be slightly hyperdense compared to the bordering brain [
17]. After application of contrast medium, in most cases a homogenous enhancement was observed [
9]. The homogenous CT-texture might reflect the histopathology and help to distinguish other malignant tumors with more necrotic areas. However, studies are needed to employ this imaging feature for discrimination purposes.
The present study corroborates the previous results with overall good comparable frequencies in regard of contrast media characteristics and density of CT images.
Regarding MRI, it was reported that 75.6% of MS lesions were isointense and 24.4% were hypointense on T1-weighted images [
16]. On T2-weighted images, 95.1% were hyperintense and 4.9% were isointense [
16]. In a recent study based on 28 patients, on T2-weighted images most cases (82.1%) were hyperintense, whereas on T1-weighted images 60.7% of the identified cases were isointense compared to adjacent muscle tissue [
9]. Similar results were also reported in another case series [
17].
Our results based on overall 54 lesions suggested similar frequencies in regard of signal intensities.
A widely used imaging technique is DWI, which can quantify proton movement in tissues and is, therefore, able to reflect microstructure of tissues [
18]. It was previously reported that DWI is very sensitive for lymphoma lesions, which show a comparable histopathology characteristic as MS lesions [
19]. So, it was reported that most MS lesions showed a diffusion restriction with up to 96% of cerebral lesions [
10]. Similar results were identified based upon 10 cases with a more advanced imaging protocol [
20]. Moreover, the ADC value increases after therapy [
20]. In the present study, the mean ADC value was slightly higher than reported, yet with most a diffusion restriction pattern. This is most likely caused by a high cellularity of the lesions, as it was extensively investigated that ADC values are inversely correlated with tissue cellularity [
18]. DWI might be a useful diagnostic tool for treatment evaluation of MS, which needs, however, more data. In regard of differential diagnosis, however, other malignant tumors and abscesses can also show restricted diffusion, which might reduce the diagnostic value of this sequence in clinical routine [
21].
A beneficial imaging modality is FDG-PET/CT, which has a better accuracy than CT in diagnosing MS [
22‐
24]. As reported previously, MS lesions show typically an intense FDG uptake. More interestingly, the tracer uptake changed under therapy, which correlated with clinical outcome [
22‐
24]. Moreover, FDG-PET can detect additional lesions, which were not clinically known [
24]. Yet, systematic data are still needed for this imaging modality to prove possible benefit. Only few lesions in the present study were staged with PET/CT, which also showed an elevated FDG uptake in good agreement with the literature.
There are several limitations of the present study to address. Firstly, it is a retrospective study with known possible inherent bias. Secondly, the patient sample is recruited from 4 German university hospitals with possible different diagnosis and treatment regimens and imaging studies were not available for all patients. Thus, due to the multicenter design the present patient sample is relatively large. Furthermore, multicenter prospective studies are difficult to perform due to low incidence of MS. Thirdly, the imaging analysis was performed in the involved centers and no central reading was performed, which might harbor some possible differences in regard of imaging assessment. However, the reading was performed by board certified radiologists and can, therefore, be generalized for clinical routine. Fourthly, the diagnosis of MS was not histopathologically confirmed in all cases. It has been reported that the misdiagnosis rate of MS is up to 47% [
2], which might be substantial lower in the present patient sample because only clinical highly suspicious MS-lesions were included in the analysis.
The oncologist and radiologist need to keep in mind the diagnosis of MS for every unclear and new occurrent lesion in a patient with known AML, independently from the localization.