To clarify the potential malignancy of eAML and investigate prognostic predictors for this disease, we studied the pathological characteristics and clinical outcomes of 67 eAML patients. More importantly, we established a risk model for eAML patients and accurately predicted the recurrence/metastasis risk. Our model provides a novel approach for the diagnosis and early intervention of eAML, a potentially malignant disease.
Renal eAML was once considered a hamartoma18. However, in subsequent years, several studies reported the malignant behavior of renal eAML, including local recurrence, distant metastasis, and death from the disease, as described in small case reports 19–21. Nevertheless, because of the rarity of renal eAML cases, the clinicopathological characteristics, diagnosis, and clinical outcomes should be further investigated for better treatment. To evaluate the potential malignancy of this rare disease, Brimo et al. studied a series of 40 renal eAML cases and demonstrated that the presence of at least 70% atypical epithelioid cells, a mitotic count of ≥ 2 per 10 HPFs, atypical mitotic figures, and necrosis were prognostic factors for renal eAML 16. Nese et al. also performed a similar study of 41 renal eAML cases and indicated that necrosis, tuberous sclerosis complex and/or concurrent AML, carcinoma-like growth, extrarenal extension and/or the involvement of the renal vein, and tumor size > 7 cm were prognostic factors for renal eAML 17. In previous studies, many predictors of poor clinical outcomes were investigated but not unified.
Studies have shown that eAML is a potentially malignant neoplasm, with approximately 30% of cases exhibiting distant metastasis to lymph nodes, liver, lungs, and spine 22. Similarly, there were 17 (25.6%) patients with metastasis in our cohort. Notably, Antonio et al. first reported a case of primary eAML of the adrenal gland in patients without evidence of tuberous sclerosis 23. Therefore, because of its similarity upon imaging, eAML is easily misdiagnosed as renal cell carcinoma or sarcoma 24. Furthermore, abnormal blood vessels and mature adipocytes are not obvious in eAML, and thus it is difficult to diagnose eAML using radiographic devices 25. Hence, the diagnosis of eAML should be mainly based on the different proportions of epithelioid components as well as positive staining of HMB-45 or Melan-A. In our eAML study cohort, all patients were diagnosed with the pathological characteristics described above, and 57 of these patients were examined to investigate the potential value of IHC indexes.
Ki-67 was first identified as an antigen by Gerdes et al. in Hodgkin lymphoma cell nuclei 26. Previous studies based on cell cycle analyses have illustrated that among the identified cell cycle markers, only Ki-67 is downregulated in the quiescent G0 phase while being highly expressed in the G1, S, and G2 phases 27In recent years, similar investigations have been undertaken in renal eAML, but the results were inconclusive. Ooi et al. demonstrated that Ki-67 was strongly positive in two eAML cases but negative in four classic AMLs 28. Moreover, Xu et al. evaluated the use of Ki-67 as a prognostic predictor in six eAML patients and found that patients with positive expression of Ki-67 had a poorer prognosis 29. Conversely, no significant difference in Ki-67 was identified between classic AML and eAML cases 30. However, all these studies were based on small sample sizes, and hence it is necessary to further verify the role of Ki-67 as a prognostic indicator in eAML patients.
The advantage of our study is that we systemically analyzed clinicopathological and IHC features and clinical outcomes in an unprecedented number of renal eAML cases. Through this, we demonstrated that renal eAML patients with malignancy exhibited larger tumor size, advanced T/N stage, and necrosis. Immunohistochemically, we also revealed that negative SMA expression and Ki-67 ≥ 10% were significantly correlated with malignancy. Based on these features, we attempted to investigate the prognostic factors of renal eAML patients and found that Ki-67 ≥ 10% and negative SMA expression were correlated with poor PFS and OS in eAML patients. Next, we established a risk model for renal eAML, which incorporated the most common, representative and accurate pathological parameters as risk factors.
This study has some limitations. First, due to the nature of the retrospective study and the missing data of some patients, we cannot but accept all the biases of our study. Second, our study did not examine the underlying mechanism of Ki-67 and SMA in the tumor metabolism of renal eAML. Third, our risk model should be verified in other renal eAML cases.