OO can be divided into three types according to its location in the bone: cortical, periosteal, and medullary. Cortical type is the most common and considered classic OO. There were 71 cases of cortical type in our study group (Fig. 1a). Osteolytic tumor nests are located in the bone cortex, surrounded by different degrees of thickening and sclerosing bone, and calcification can be found in the nests. Removal of the tumor might result in a significant decrease in bone strength. Periosteal tumor nests are located under the periosteum, adjacent to the bone cortex, with accompanying periosteal reaction. Removal of the tumor has little effect on bone strength. There were 20 cases of periosteal type in our study group (Fig. 1b). Medullary tumor nests occur in the cancellous bone of the medullary cavity, often in the neck of femoral head and vertebral body, and there is no obvious hyperosteogeny and sclerosis around the tumor nests [17, 18]. We needed to open a window in the cortical bone to remove the tumor. Fortunately, the fenestrated bone can be retained and replanted. There were 3 cases of the medullary type in our study group (Fig. 1c). OO tumor cells produces excessive prostaglandin, which causes widespread bone pain. Therefore, complete pain relief in patients is a sign of complete tumor resection .
At present, surgical methods for in situ inactivation treatment of OO have become popular, such as radiofrequency ablation, freezing, and radiotherapy. Radiofrequency ablation is the most recommended method because of the advantages of minimal surgical trauma, short operation time, and short hospital stay . However, the shortcomings of this technique need to be considered. The mechanism of radiofrequency ablation is by thermal destruction of the tumor tissue. It is difficult to accurately control the degree and range of heating during the process, therefore, it is easy to cause damage to important surrounding tissue structures and neurovascular bundles, unlike in an open excision surgery. By performing precise open excision surgery such as in our study, only tumor nests are removed and normal bone loss is minimized. Comparatively, radiofrequency ablation, freezing, or radiation therapy cannot be so precise. In order to ensure effective tumor cell killing, tissue is often included outside the tumor nest, resulting in more damage to the bone and periosteum [20, 21]. Therefore, precise open excision surgery has lesser bone damage and loss than the techniques of tumor in situ inactivaton. The recovery in case of bone injury is very slow, which is a key factor for patients to recover and resume their daily activities. Open surgery does cause more soft tissue damage, but this can recover well in two to three weeks. Moreover, since we expose from the intermuscular space, it causes minimal damage to the soft tissue structures. Oc Yunus et al.  applied the radiofrequency method to 58 patients and complications were observed in seven patients. Second-degree burns were seen in two patients, and superficial skin infection developed in two patients. In one patient, the probe tip broke and remained within the bone. Intramuscular hematoma was detected in one lesion located in the proximal femur. There was another case of numbness of fingers after the operation, suspected to be caused due to nerve damage. Lassalle et al.  additionally reported two cases of peripheral neuropathy after surgery, one case with arm neuropathy from nerve damage, one case with electrode rupture, and one case with muscle hematoma. None of the above complications occurred in our study.
In terms of speed of patient recovery, precise open excision surgery also has advantages. For periosteal OO, there was no significant change in bone strength after resection. Comparatively, in situ inactivation can cause widespread peripheral soft tissue damage and bone cortex damage, with minimal patient benefit. For OO located in the cortex, following both resection or in-situ inactivation, the bone strength decreases, which may cause a pathological fracture. However, in an open surgery, the loss of bone cortex is smaller, and it can be fully implanted or assisted with internal fixation, decreasing the risk of a pathological fracture. Furthermore, in situ inactivation usually results in widespread cortical necrosis, and therefore, cannot effectively increase the bone strength, resulting in higher incidence of fractures. For medullary OO, it should be the most suitable way for tumor inactivation in situ. However, there are not many such patients. The periosteum and cortex can be preserved when the window of bone is opened by open precise resection. After the tumor is removed, the bone cortex can be replanted, and the healing is very fast. In other studies, pathological fracture and limited bone union were the main complications of the surgical treatment of OO, with an incidence rate of up to 8% [9, 24]. Ruiz et al.  reported that among 26 patients with bone tumors treated by radiofrequency ablation, one had symptomatic bone infarction, and another one had a pathological fracture. None of these complications were observed in our study, which additionally confirms the effectiveness of our bone grafting method.
OO as an osteogenic tumor and it usually needs to be differentiated from osteoblastoma, chronic osteomyelitis, and even osteosarcoma . Therefore, the definite diagnosis of OO is significant for deciding the treatment plan and for the estimation of prognosis. However, imaging examination has some limitations in the diagnosis of OO, as it completely depends on the degree of osteosclerosis and periosteal hyperplasia of OO. For OO that occurs under the periosteum or on the surface of the cortex, the degree of sclerosis is relatively light, with periosteal reaction. For OO that occurs in the marrow cavity, because it is away from the inner and outer periosteum, periosteal hyperplasia rarely occurs, and the degree of sclerosis is also relatively light. "Bull's eye sign" is not typical; therefore, the imaging examination is prone to misdiagnosis [9–11]. Consequently, histopathological examination is essential to confirm the diagnosis, and open surgery can obtain pathological specimens to make a diagnosis. In contrast, in situ inactivation cannot obtain effective pathological samples, increasing the risk of misdiagnosis and missed diagnosis [16, 26]. The imaging examination might suggest probable osteoid osteoma in some cases, and the final pathological result reveals osteomyelitis or even osteosarcoma, which was also observed in other studies. Singh et al.  reported six cases that were initially misdiagnosed as tuberculous arthritis or osteomyelitis, and finally confirmed as osteoid osteoma by pathological examination. Similarly, Georgiev et al.  reported a 12-year-old boy with pain and swelling of the middle phalanx of the ring finger that did not respond to anti-inflammatory drugs for three months. The imaging data was suspicious of osteomyelitis or Ewing's sarcoma, but the pathological examination confirmed osteoid osteoma. Therefore, pathological examination is necessary for the diagnosis of OO, and since all of our patients had been diagnosed with OO by pathological examination, there was no misdiagnosis or missed diagnosis.
Since the recurrence rate is still the key index to judge the efficacy of bone tumor surgery [29–31], our patients were followed up for a mean of 36.9 months. In case of 93 patients, there was no recurrence of pain, and no recurrence was found on X-ray reexamination. Only one patient (1.1%) was found to have suspected recurrence 50 months after the first surgery, but the location of tumor recurrence was not consistent with the original tumor location. In addition, there had been no pain during the follow-up period of 50 months. The patient remained free of any discomfort after the second operation. Therefore, this patient was considered to have a possible new tumor, rather than a recurrence, but the location was on the same bone. It has not been fully ascertained whether this was a recurrence or a primary tumor, and remains open to discussion. However, in studies on in situ inactivation, the results were not satisfactory. Lindler et al.  utilized the radiofrequency technique in 58 patients and reported a recurrence on months 3, 5, and 7, respectively, in three patients (5.2%). Baal Joe et al.  reported 71 cases of CT-led radiofrequency ablation therapy, with 10/71 patients (14%) experiencing a symptomatic relapse on an average 21.5 months after surgery. Therefore, the recurrence rate in our study group (1.1%) was significantly lower than reported in other related studies.
Essentially, we compared the precise open excision surgery technique with the in-situ inactivation operation in terms of trauma, bone strength, bone healing, diagnosis, and recurrence rate, and found that the precise open excision surgery is indeed a more reliable choice. On the basis of our experience, we would recommend paying close attention to the following points in the treatment process in order to ensure maximum patient benefit: diagnosis of disease, no detour, no misdiagnosis, no missed diagnosis, clear focus, no deception by imaging, precise treatment, direct to the focus, clear it thoroughly under direct vision, and be careful of recurrence.
Our study had certain limitations. First, we had a relatively short follow-up period for patients with a secondary surgery. Second, intraoperative navigation can be more accurate to find the focus and further reduce the damage. Lastly, our study was mainly retrospective and lacked a direct comparison group or randomized control. However, the overall complication and recurrence rate of our study was lower compared to other reports. All the patients in the study had a great outcome.