This retrospective review of multimodality imaging in chondroblastomas reiterates its ‘unlike-cartilaginous lesion’ imaging appearance. Our study cohort suggesting male (3:1) preponderance with mean patient age being 18 years (range: 10–57 years) was in line with existing literature [1,2,3]. As described in multiple previous studies, 75% of our cases were observed in immature skeletons with 46 cases involving long bones showing a striking epi/apophyseal affinity (n = 46). Most common site in our study (proximal humerus) was similar to findings of various previous studies [1,2,3, 9, 10]. Atypical distribution was frequent in older patients, with patients older than 30 years of age showing involvement of scapula (n = 1), lunate (n = 1), acetabulum (n = 1), skull base (n = 1), and temporal bone (n = 2). This is similar to findings of Bomhauer et al. where atypical short tubular and flat bone involvement [4], higher metaphyseal involvement, and relatively more aggressive behavior in terms of rates of local recurrence and metastases were observed with increasing age. About 47% (n = 22) of our appendicular cases were found to have subarticular extension which conventionally has been cited as a classic feature of giant cell tumors (GCT). However, more than half of these cases were non-eccentric, a feature that may aid in morphologic distinction from GCT.

Type of margin and cortical breach were better visualized on CT which was similar to the observations of Weatherall et al. The presence of a characteristic sclerotic rim can help in distinguishing CB from clear cell chondrosarcomas which tend to permeate the outer cortex [11] and GCT which usually lack a sclerotic rim apart from their propensity to afflict older patients and lack of perilesional edema incitement.

Dia-metaphyseal solid or layering periostitis away from the lesion described by Brower et al. [12] in 47% of their cases (101/214) was observed in ~ 15% cases (n = 8) in our study. Periosteal reaction, when present, was thick and solid which can hint toward the benignity of this aggressive appearing entity (Fig. 3).

On MRI, this ‘dark horse’ of cartilaginous lesions demonstrated a predominantly low T2WI signal intensity with high SI foci within, in 66% of our cases (n = 34). This is in line with the findings of Weatherall et al. who found this characteristic appearance in 16 of their 22 cases (73%) [7] and Jee et al. who observed it in 10 of their 22 cases (45%) [13]. They further found that higher percentage of high SI foci within a lesion corresponded with higher percentage of lakes of hyaline cartilage and/or hemorrhage on histology. Such correlation was not possible in our study as curettage specimens were available in only 20 cases and only representative tissue was sent for histologic evaluation in 32 cases that were treated by radiofrequency (n = 31) or cryoablation (n = 1). Combination of highly cellular chondroid matrix and calcification has been postulated to cause the decrease in T2 relaxation time [7, 14, 15].

Touted to be a cardinal feature associated with CB, perilesional edema was found in 94% (n = 49) cases with moderate to severe (equal to more than the lesion diameter) edema in 44% cases (n = 23), often seen extending to adjacent muscles and inciting joint effusion in 50% cases (n = 26). J.W et al. encountered joint effusion in 75% of their cases, majority due to direct joint invasion by the lesion with few cases of effusion in the absence of direct joint involvement [14].

Perilesional alterations appearing as subtly increased intramedullary density on radiograph, hypointensity on T1W, and hyperintensity on STIR/T2W MRI images, with radiopharmaceutical uptake on metabolic imaging associated with CB, play a part in its sinister appearance. Locally acting enzymes released by blastoma cells and bone morphogenic proteins (BMPs) have often been held culprit for the peritumoral edema. Weatherall et al. felt this characteristic ancillary finding could not be explained by Frost’s ‘regional acceleratory phenomenon’ or Bower et al.’s ‘stress phenomenon’ alone, though it could be an analog of the ‘flare phenomenon’ described for osteoblastomas by Crim et al. [16, 17].

Another feature lending it an aggressive appearance included associated pathological fractures (n = 2); however, they are likely to be exceptions than the rule [6, 18]. Secondary, aneurysmal bone cyst causing fluid–fluid levels have been seen in up to 15% cases of CB; however, we encountered only two cases (acetabulum, femur) showing such an appearance in our cohort.

Like Jee et al. we observed enhancement in all cases, demonstrating either heterogeneous lobular or septal pattern [13]. Dynamic enhancement curves were available for two cases and were of type 2 (plateau), further pointing toward benign etiology (Fig. 1). 50% of the cases showed restricted diffusion (Fig. 2) which was patchy more often than not.

Post-treatment MRI showed increased lesion heterogeneity on T2 and T1 images and continued to show persistent minimal to mild enhancement which was stable or decreased as compared to pre-treatment scan. There was near-complete resolution of perilesional edema on MRI and no radionuclide uptake in the lesion seen on PET/CT which served as the predominant marker for complete tumor ablation (Fig. 2).

We acknowledge the potential bias in the study due to its retrospective nature. Pathologic correlation for specific imaging findings was limited as curettage was performed in only a few cases due to evolving management strategies for benign bone tumors. Finally, there might have been variations in image acquisitions across different scanners; however; non-standardized imaging protocols can perhaps be viewed as a plus, since it suggests that results may be generalized and will likely be less dependent on protocol variations between centers/scanners.

In conclusion, chondroblastoma has distinctive imaging appearance and is often unlike majority other cartilaginous benign lesions due to characteristic low T2 signal on MRI and associated exuberant perilesional edema (Table 1).