Infantile BCG osteomyelitis ultimately diagnosed as a result of trauma

Abstract

The BCG vaccine is effective in preventing severe tuberculosis infections in infants and young children, and it is commonly administered in Japan to those aged between 5 to 8 months. Although the incidence of adverse reactions is low, disseminated lymphadenitis and osteomyelitis have been reported as potential side effects. We encountered a case of BCG osteomyelitis with typical imaging findings. Provided the clinical presentation and imaging features are recognized, it is possible to diagnose BCG osteomyelitis definitively and exclude a number of possible differential diagnoses.

Introduction

Bacille Calmette-Guerin (BCG) vaccine prevents severe tuberculosis infections such as disseminated tuberculosis in infants and young children. Administering it to children under the age of 1 reduces tuberculosis-related mortality by 71% [1]. Adverse reactions following BCG vaccination include suppurative lymphadenitis, disseminated BCG infection, osteomyelitis, anaphylaxis, and skin tuberculosis-like lesions [2]. Unlike typical bacterial suppurative osteomyelitis, BCG-induced osteomyelitis presents with milder systemic symptoms such as fever and localized pain, and blood biochemical findings rarely indicate acute inflammation [3]. Additionally, plain X-rays often show bone destruction extending from the metaphysis to the epiphysis, leading to suspicion of bone malignancies. In this report, we describe a case of post-BCG osteomyelitis with a typical clinical course and plain X-ray findings, focusing on its characteristic imaging features and differential diagnosis.

Case presentation

The patient was a 16-month-old boy with no significant medical or family history. After falling on a hardwood floor at home, he began to complain of right shoulder pain. He visited a nearby orthopedic clinic, where plain X-rays were taken. The results suggested a bone tumor or fracture of the right humerus, leading to a referral to our hospital. At initial presentation, there was no fever or localized pain, and blood tests showed that the levels of inflammatory markers were low.

A plain X-ray of the right shoulder joint revealed an irregular radiolucent area extending from the proximal diaphysis to the metaphysis of the right humerus (Fig. 1). The continuity of the cortical bone on the axillary side was lost, showing destructive deformation. These findings excluded a simple fracture due to trauma as initially suspected by the referring physician and revealed the following characteristic features:

Fig. 1.

Plain X-ray of the right shoulder joint (frontal view). A continuous radiolucent area is observed in the proximal metaphysis and epiphysis of the right humerus (arrow). The internal structure of the radiolucent area is heterogeneous, with multiple linear structures suggesting that it is not a simple cystic lesion. The continuity of the cortical bone is lost, showing destructive deformation. A linear periosteal reaction is present, but there are no malignant indicators such as Codman's triangle, spiculation, or sunburst appearance. The radiolucent area has clear, circular margins with no permeative lesions or “moth-eaten” appearance. There is no evidence of expansive growth.

The circular radiolucent lesion at the proximal humeral metaphysis had clear margins with a sclerotic border, suggesting that the growth rate was not rapid. Additionally, the internal structure of the radiolucent area showed linear components, suggesting it was not a simple cystic lesion. The proximal humeral cortex displayed a linear periosteal reaction, but there were no malignant periosteal reactions such as Codman's triangle, spiculation, or sunburst appearance, ruling out malignancy based on these findings. Furthermore, there were no permeative lesions and the diaphysis did not have a “moth-eaten” appearance, suggesting that it was not a typical malignant tumor. The morphology of the lesion was nonexpansive, with no increase in the transverse diameter of the diaphysis.

Additionally, a radiolucent area was observed in the center of the humeral epiphysis (Fig. 1, arrow). The lesions in the epiphysis also appeared to have clear margins and lacked destructive changes suggestive of malignancy. This was considered to be an infiltration from the primary lesion in the diaphysis and metaphysis across the growth plate into the epiphysis.

Thus, plain X-ray ruled out the possibility of primary malignant bone tumors or simple fractures. Inflammatory diseases, particularly osteomyelitis of the humerus, were considered as the top differential diagnosis. However, the absence of fever, pain, and low levels of inflammatory markers in blood tests, along with the presence of tumor-like radiolucent areas and bone destruction, were atypical for distinctive bacterial osteomyelitis. Therefore, we suspected osteomyelitis due to specific infections such as tuberculosis. The differential diagnosis includes the following diseases: simple bone cyst [4], fibrous dysplasia [5], aneurysmal bone cyst [6], chondroblastoma [7], Langerhans cell histiocytosis (LCH) [8], and giant cell tumor [9]. Although primary malignant bone tumors such as osteosarcoma and Ewing's sarcoma were less likely due to the patient's age, metastatic bone tumors such as neuroblastoma were considered. Other differential diagnoses included chronic nonbacterial osteomyelitis (CNO) [10] and kaposiform hemangioendothelioma (KHE) [11,12], an infiltrative vascular anomaly.

Contrast-enhanced CT (not shown) subsequently showed destructive changes extending from the diaphysis to the epiphysis of the right humerus. No primary malignant tumors such as neuroblastoma, which could cause distant metastasis, were detected. Nor was there any significant lymph node swelling. Right shoulder joint MRI revealed a high signal area extending from the metaphysis to the epiphyseal marrow on T2-weighted images (STIR) similar to the findings on plain X-ray (Fig. 2A, arrow). The lesion also extended into the soft tissues beyond the cortical bone (Fig. 2A, arrowhead). Diffusion-weighted imaging showed a high signal with restricted diffusion (Fig. 2B, arrow), but the ADC value on the ADC map was about 1.1 × 10^-3 mm²/sec, not typical findings for primary or metastatic malignant lesions or abscesses (Fig. 2C, arrow). No early enhancement was observed on contrast-enhanced MRI, but delayed enhancement was noted in the bone marrow and around the lesion (Fig. 2D, arrow). There was no joint effusion or other abnormalities within the scan range. No signs of metastasis or multiple bone lesions were evident on bone scintigraphy (not shown). These findings were suggestive of tuberculous osteomyelitis.

Fig. 2.

MRI of the right shoulder joint. T2-weighted image (STIR) (A) shows a high signal area in the bone marrow extending from the metaphysis to the epiphysis of the humerus (arrow). High signal intensity is also seen in the soft tissues beyond the cortical bone, suggesting the spread of inflammation (arrowhead). Diffusion-weighted image (B) shows a high signal corresponding to the lesion (arrow). On the ADC map (C), there is a signal drop in the lesion area, but the ADC value is approximately 1.1 × 10⁻³ mm²/sec, which is not particularly low. The fat-suppressed T1-weighted image after gadolinium enhancement (D) shows heterogeneous enhancement in the bone marrow.

A bone biopsy was performed. Pathologically, hematoxylin-eosin (HE) staining revealed multiple epithelioid granulomas of various sizes, including some with necrosis. (Fig. 3). Acid-fast bacilli were not detected by Ziehl-Neelsen staining. Subsequently, DNA was extracted from the bacterial strain at the Tuberculosis Research Institute, and PCR identified Mycobacterium bovis BCG.

Fig. 3.

Pathological diagnosis revealed epithelioid granuloma with necrosis on hematoxylin eosin (HE) staining.

This case was diagnosed as BCG osteomyelitis, as the patient had received BCG vaccination in the left upper arm at age 5 months, 1 year prior to onset.

Interferon gamma receptor 1 (IFNGR1) expression, tested to assess susceptibility to mycobacteria including tuberculous/nontuberculous mycobacteria, was normal, and no pathogenic variants were found in the Mendelian susceptibility to mycobacterial diseases (MSMD) panel. Chronic granulomatous disease was ruled out due to normal reactive oxygen species production by neutrophils.

The patient was treated with isoniazid and rifampicin according to the tuberculosis treatment guidelines and his condition improved.

Discussion

BCG osteomyelitis is a low-grade inflammatory disease in infants, with an incidence rate of 0.01 to 300 per million BCG vaccinations [13]. BCG osteomyelitis is most often diagnosed 6 to 12 months after vaccination, but cases diagnosed as long as 12 years postvaccination have also been reported [14]. The most common sites are the long bones of the extremities, but it can also be seen in the tarsal bones, metatarsal bones, sternum, ribs, spine, ilium, and clavicle. Unlike tuberculosis, BCG osteomyelitis rarely involves arthritis or spondylitis. Although the preferred sites differ from those of tuberculosis, the imaging findings are similar. In long bones, it typically involves the epiphysis or metaphysis, and as in this case, it can extend through the growth plate. Surrounding bone sclerosis is minimal or absent, and periosteal reaction is weak. Bone necrosis is not observed [15].

Although BCG osteomyelitis is considered to have a good prognosis, early diagnosis and treatment is desirable because recurrence and sequelae may occur if the infection spreads beyond the growth plates [15]. Ohtera et al. reported a case of BCG osteomyelitis involving the growth plates that was treated by orthopedic surgery but did not subsequently cause growth and developmental disorders [16].

The differential diagnosis includes bacterial pyogenic osteomyelitis; however, the clinical presentation of typical pyogenic osteomyelitis is usually more pronounced. In contrast, tuberculous osteomyelitis and BCG osteomyelitis generally lack systemic symptoms such as fever and local signs such as redness and pain. Additionally, bloods tests generally reveal only a mild increase in inflammatory markers [15].

Based on the characteristic imaging findings of this case, differential diagnoses were considered. Generally, the mechanism by which infection spreads from the metaphysis to the epiphysis is understood as follows: In infancy up to 18 months, blood flow from the metaphysis to the epiphysis is observed through trans-physeal capillaries across the growth plate. However, in toddlers and school-age children up to 18 years, this blood flow ceases with the closure of the growth plate. As adulthood is reached, the blood flow resumes [[17], [18], [19]]. In this case, since the growth plate had not yet closed, it was possible for the inflammation to extend from the metaphysis to the epiphysis even in bacterial osteomyelitis.

Tsujioka et al. described the characteristic finding of BCG osteomyelitis as the destruction of the growth plate and spread to the epiphysis by the infected nests at the metaphyseal end, and reported that this was seen in 82% of cases [3]. Another characteristic of BCG osteomyelitis, as compared to bacterial osteomyelitis, is the presentation of coarse bone radiolucent areas in the metaphysis, which is consistent with this case [3]. Some children who develop BCG osteomyelitis may have a congenital susceptibility to BCG, known as Mendelian susceptibility to mycobacterial diseases (MSMD). MSMD includes the following heterogeneous groups: genetic diseases associated with autosomal dominant or sporadic interferon-gamma-receptor (IFN-γ-R) deficiency [3]. Although MSMD was suspected in this case, it was not consistent with the typical features of MSMD, which include a median onset of 5 months, multiple lesions, and severe clinical presentation [3,20]. Additionally, MSMD is characterized by lymphadenopathy in the early stages of the disease, which was not observed in this case, further suggesting that MSMD was unlikely [3].

Next, it is well known that bone tumors have a characteristic age of onset and common sites. In terms of the types of lesions occurring in the diaphysis and metaphysis, simple bone cysts and fibrous dysplasia occur frequently, but it is rare for these lesions to extend beyond the growth plate to the epiphysis. Among tumorous lesions that extend from the diaphysis to the epiphysis, chondroblastoma is a differential diagnosis; however, it most commonly occurs in individuals with an average age of 19-23 years [7], and it seldom causes cortical destruction [7], which is different from the findings in this case. Additionally, giant cell tumors, which can extend to the epiphysis, predominantly occur in young adults [9], making it less likely in this patient's case considering his age.

Langerhans cell histiocytosis (LCH) occurs between the ages of 1 and 15, with a peak incidence at 3 years of age [8], with the long bones being a common site. Zhang et al. studied 22 cases of LCH in long bones and reported that the characteristic findings on plain X-rays included lytic lesions (100%), soft tissue swelling (95%), periosteal reaction (77%), expansile appearance (70%), and cortical destruction (50%) [8], all of which were observed in this case. Therefore, LCH cannot be excluded from the differential diagnosis.

Chronic nonbacterial osteomyelitis (CNO) typically presents with bone pain (97%), nocturnal bone pain (80%), and loss of function (73%) as reported by Kaut et al. [10], which does not align with this case. Vascular anomalies such as kaposiform hemangioendothelioma (KHE) are considered unlikely due to the absence of abnormal vascular findings on Gd-enhanced imaging [12].

Conclusion

When clinical symptoms are not obvious and osteomyelitis extending from the metaphysis to the epiphysis is diagnosed, tuberculous osteomyelitis should be included in the differential diagnosis. Especially in children who are aged 6 to 12 months after receiving BCG vaccination, BCG osteomyelitis should be kept in mind. Early diagnosis of BCG osteomyelitis is clinically difficult, and as mentioned above, the differential diagnosis is wide-ranging and diagnosis may be delayed. However, these diseases can be excluded by careful interpretation of plain X-rays and MRI images, and it appears that this is a disease in which diagnostic radiologists can contribute to early diagnosis.

Author contributions

Katsuhiko Tabata and Osamu Miyazaki conceived the idea for the article. Katsuhiko Tabata and Osamu Miyazaki wrote the main body of article. Shunsuke Nosaka supervised the whole article. Ayako Imai, Reiko Okamoto, Mikiko Miyasaka, and Yoshiyuki Tsutsumi performed the literature search and data analysis. Yoko Anami and Atsuhito Seki obtained samples during surgery. Tatsuki Ikuse, Toshinao Kawai, and Takashi Ishikawa collected the data. Takako Yoshioka and Chizuko Haga prepared pathological specimens and made the diagnosis. All authors discussed the results and commented on the manuscript.

Patient consent

Informed consent for publication was obtained from the parents of the patient.

Since the pediatric patient was very young and lacked decision-making capacity, consent was obtained from the parents.