Does PET/CT Aid in Detecting Primary Carcinoma in Patients with Skeletal Metastases of Unknown Primary? : Clinical Orthopaedics and Related Research®

Secondary Logo

Journal Logo


Does PET/CT Aid in Detecting Primary Carcinoma in Patients with Skeletal Metastases of Unknown Primary?

Lawrenz, Joshua M. MD; Gordon, Jaymeson; George, Jaiben MBBS; Haben, Collin DO; Rubin, Brian P. MD, PhD; Ilaslan, Hakan MD; Mesko, Nathan W. MD; Nystrom, Lukas M. MD

Author Information
Clinical Orthopaedics and Related Research 478(11):p 2451-2457, November 2020. | DOI: 10.1097/CORR.0000000000001241



Metastatic carcinoma of bone is the most common osseous malignancy in adults and is most often secondary to primary breast or prostate cancer [4]. In 1993, Rougraff et al. [6] suggested a diagnostic algorithm to identify the primary cancer in patients presenting with a radiolucent bone lesion that is likely to be a skeletal metastasis but no known primary cancer. This diagnostic evaluation included a history and physical examination, laboratory tests, extremity radiographs, chest radiographs, whole-body bone scan, CT of the chest, abdomen, and pelvis, and biopsy of the most accessible lesion. This evaluation was shown to be 85% successful in diagnosing the malignancy. Since that seminal work, other investigators have reported similar diagnostic yield [2, 3, 8, 9]. Although this is indeed excellent, this leaves 15% of patients with a primary cancer of unknown origin, labeled as a skeletal metastasis of unknown primary. Survival times are known to be markedly decreased in the setting of carcinoma of unknown primary [2-4, 7, 8]. As targeted therapies evolve, obtaining the correct diagnosis is imperative to assigning the appropriate treatment and maximizing survival.

PET/CT first became commercially available to medical centers in 2001 and has since assumed a prominent role in cancer staging and surveillance [5]. In the setting of unknown primary cancer, PET/CT has been proposed to have value in identifying pathologic metabolic changes in the absence of anatomic changes not found by traditional advanced imaging (whole-body bone scan and CT of the chest, abdomen, and pelvis) [1, 5]. However, clinical data using PET/CT in this manner are limited, with heterogeneous patient cohorts and inconclusive results. An investigation in 2012 reported that PET/CT identified the primary cancer in 43% (54 of 120) of patients with a (primarily soft tissue) metastasis of unknown primary not identified with the standard diagnostic algorithm, and advocated that PET/CT should be considered earlier in the evaluation of these patients [1]. This investigation was limited because it primarily evaluated non-osseous metastatic sites, as only 20% (24 of 120) of the patients had a skeletal metastasis. It therefore has limited applicability to the physician evaluating a potential metastatic bone lesion. Two study groups in Japan recently stated that there was no statistical difference between PET/CT and traditional CT in detecting primary lesions [7], and recommended CT alone for its relatively cheaper cost and shortened wait time [7, 8]. Given the opposing nature of these results and the limitations cited, our purpose was to assess the diagnostic performance of PET/CT in patients with a skeletal metastasis of unknown primary.

Specifically, we aimed to answer two questions: (1) What is the ability of PET/CT to detect the source of the primary tumor in patients with a skeletal metastasis of unknown primary? (2) How does PET/CT perform in detecting metastases in other sites in patients with a skeletal metastasis of unknown primary?

Patients and Methods

After internal institutional review board approval, we retrospectively analyzed the pathology database at our institution between January 2006 and December 2016. Patients with both a histologically confirmed bone carcinoma and a PET/CT result in their medical record were identified (n = 182). Patients younger than 18 years and those who had a non-spinal skeletal metastasis of the head and neck were excluded from this search. The following additional exclusion criteria were applied: (1) patients with an incomplete standard diagnostic evaluation, which consisted of a history and physical examination, laboratory evaluation, CT of the chest, abdomen, and pelvis, and bone metastasis biopsy (n = 70); (2) patients who underwent PET/CT before the standard diagnostic evaluation (n = 21) because we felt this would affect the read of their subsequent standard imaging studies; (3) patients evaluated with PET/CT more than 3 months after identification of a skeletal metastasis (n = 37); and (4) patients who underwent resection of the primary cancer before the presentation of a skeletal metastasis (n = 19). There were 35 patients, with 176 total metastatic sites, who met the inclusion criteria for this study (Fig. 1).

Fig. 1:
This flowchart shows the patient inclusion and exclusion criteria. aPatients younger than 18 years or those who had a skeletal metastasis of the head or neck were excluded from this search.

We included only patients who had a PET/CT performed within 3 months after the standard diagnostic evaluation. The indications for a PET/CT were not limited. They primarily consisted of patients who had the examination ordered for further evaluation of a potential primary site, or those who underwent the examination as part of the standard staging process for the newly diagnosed malignancy (such as, non-small cell lung carcinoma). Regardless of the indication for the PET/CT, the patient was included.

We manually extracted basic demographic information including patient age, sex, and location of the skeletal metastatic lesions from our electronic medical record (Table 1). Among the patients who qualified for this study, the standard diagnostic evaluation identified the primary cancer in 22 of 35 patients, most commonly with a CT of the chest, abdomen, and pelvis (68%; 15 of 22 patients). Patients with a known primary had a mean age of 63 years ± 14 years; 55% were men (12 of 22) were men and 45% were women (10 of 22). Patients with a skeletal metastasis of unknown primary had a mean age of 63 years ± 11 years; 10 of 13 were men and three of 13 were women (Fig. 2). The location of the metastatic lesions was described as the upper extremity, lower extremity, pelvis, spine, or chest wall. The 35 patients included in the study were further divided for analysis into two groups: patients with a skeletal metastasis of unknown primary and those who had a known primary cancer, based on whether the standard diagnostic evaluation identified the primary cancer. Thirteen patients had a skeletal metastasis of unknown primary at the time of the PET/CT evaluation. We evaluated the diagnostic accuracy of PET/CT in detecting the unknown primary cancer. For the 22 patients in whom the standard diagnostic algorithm identified the primary cancer (known primary cancer group), the accuracy of PET/CT in confirming the primary cancer site was assessed by noting whether the PET/CT scan showed increased uptake at the site of the known primary cancer, consistent with malignancy.

Table 1. - Diagnostic ability of PET/CT to identify the primary cancer
Demographic/tumor characteristic Number of patients with unknown primary cancer at the time of PET/CT (n = 13) Number of patients with known primary cancer at the time of PET/CT (n = 22)
Age (years), mean ± SD 63 ± 11 63 ± 14
 Male 10 12
 Female 3 10
Skeletal location
 Upper extremity 1 0
 Lower extremity 2 7
 Pelvis 4 7
 Spine 6 6
 Chest wall 0 2
Ability of PET/CT to detect primary cancer
PET/CT detected/confirmed primary cancer 1 22
 Lung 0 15
 Breast 0 2
 Thyroid 0 2
 Esophagus 0 1
 Laryngeal 0 1
 Oropharynx 1 1
PET/CT did not detect primary cancer location 12
 Remained unknown after PET/CT 12
Follow-up of this patient cohort was for 3 months after PET/CT to identify any missed primary or metastatic tumor.

Fig. 2:
This figure shows the diagnostic performance of PET/CT to identify or confirm the primary cancer site.

The ability of PET/CT to detect and confirm both skeletal and non-skeletal metastases was calculated for the entire cohort of 35 patients. The number and location of metastatic sites identified before imaging with PET/CT was recorded and then compared with the areas of increased uptake on subsequent PET/CT images. All metastatic sites not previously identified by evaluation before PET/CT but identified as increased areas of uptake consistent with metastases by PET/CT were recorded and defined as newly identified metastases. The numbers of confirmed and newly identified metastatic sites were summed as well as categorized by location of the metastasis (Table 2). In the entire cohort of 35 patients, 176 metastatic sites were identified with the standard imaging evaluation and PET/CT. Most metastases were of bone (55%; 97 of 176) and lymph nodes (19%; 34 of 176). The PET/CT findings of metastatic disease were not uniformly confirmed histologically or with additional radiographic imaging (such as MRI). In the patients where additional confirmation occurred within a 3 month time frame after the PET/CT, it was noted whether the area was or was not confirmed to be a likely metastatic focus. The ability of PET/CT to detect metastatic sites as compared with the standard diagnostic algorithm of bone scan and CT chest/abdomen/pelvis was quantified by calculating the false-negative and false-positive rates. Sensitivity and specificity were also calculated for PET/CT using the previously calculated false-positive and false-negative rates.

Table 2. - Diagnostic ability of PET/CT to identify metastatic sites
Metastatic site Total percentage of all 176 metastatic sites (n) Percentage of the 176 metastatic sites (n) detected by whole body bone scan and CT chest/abdomen/pelvis Percentage of the 176 total metastatic sites (n) detected by PET/CT Percentage of false negatives by PET/CT (n)a Percentage of false positives by PET/CT (n)b
All sites 100 (176) 65 (115) 35 (61) 3 (3 of 115) 7 (2 of 26)
Lung 8 (15) 93 (14) 7 (1) 0 (0) 0 (0)
Liver 1 (2) 100 (2) 0 (0) 0 (0) 0 (0)
Other intraabdominal or intrapelvic viscera 6 (10) 70 (7) 30 (3) 0 (0) 0 (0)
Brain 2 (2) 100 (2) 0 (0) 50 (1 of 2) 0 (0)
Bone 55 (97) 62 (60) 38 (37) 3 (2 of 60) 0 (0)
Soft tissue 9 (16) 62 (10) 38 (6) 0 (0) 2 (1 of 26)
Lymph node 19 (34) 59 (20) 41 (14) 0 (0) 2 (1 of 26)
aThe false-negative percentage was calculated as the number of metastatic sites missed by PET/CT, divided by the number of metastatic sites known before PET/CT.
bThe false-positive percentage was calculated as the number of new areas of uptake that PET/CT identified that could not be identified as metastases after additional radiographic imaging, divided by the number of new areas of uptake found by PET/CT that had additional radiographic imaging after PET/CT. Only 26 of 61 new areas of uptake on PET/CT had additional imaging that could assess whether the sites were identifiable metastases on radiographs.


Identification and Confirmation of Primary Cancer

PET/CT successfully confirmed the site of the primary cancer in all 22 patients who had their primary detected by the standard diagnostic algorithm. Of the remaining 13 patients classified as having a skeletal metastasis of unknown primary, PET/CT detected the primary cancer in only one of 13. The one primary cancer that was successfully identified by PET/CT arose from the oropharynx.

Identification and Confirmation of Metastatic Sites

Of the 115 metastatic sites found after the standard diagnostic evaluation, PET/CT only failed to identify three, yielding a false-negative rate of 3% (three of 115) and a sensitivity of 97% (112 of 115). Furthermore, PET/CT identified 61 new areas suspicious for metastases not detected by the standard algorithm. Of these, eight of the 61 new areas were skeletal metastases that were not detected by bone scan. Of the 61 areas of uptake on PET/CT consistent with newly identified metastases, only 26 had further radiographic imaging to confirm the PET/CT findings. The remaining 35 patients were presumed metastatic based on clinical and PET/CT findings, without additional confirmation. Of these, 24 of the 26 newly identified metastatic sites identified with PET/CT were supported with further radiologic evaluation, yielding a false positive rate of 8% (two of 26) and a positive predictive value of 92% (24 of 26). The other two patients were determined to have benign chondroid neoplasms on MRI evaluation.


Over the past 25 years, little improvement has been made in the diagnostic evaluation of patients with a skeletal metastasis from unknown primary. The aim of this study was to assess the utility of PET/CT imaging in identifying the primary cancer in the 15% of patients whose cancer is unidentified after the standard diagnostic algorithm [6]. Our data show that PET/CT imaging was rarely able to locate a primary cancer in this subset of patients. However, in patients whose cancer was detected with the standard diagnostic algorithm, PET/CT consistently confirmed the site of the primary cancer and known metastatic sites. PET/CT was also particularly useful in identifying new metastatic sites, primarily of bone, suggesting it could be useful as a screening tool for osseous metastatic burden.

Our investigation has several limitations. As a retrospective study, we were dependent largely on the accuracy of radiographic reads and diligent image storage to compile our data. Although a limitation, data of this nature are less subject to bias given that most were taken from actual radiographic images or radiographic reports. The manner in which patients were identified for this analysis also means that our study only included patients who had a PET/CT. This methodology yielded a low number of patients who had evaluation with a PET/CT aimed at detecting a truly unknown primary neoplasm. Given this low number events, the results should be interpreted with caution. This was a retrospective study, which means that the PET/CT was ordered either as an additional test to identify the primary source or for staging of their neoplasm. This does not discredit the findings from our report because regardless of the rationale for the PET/CT, the objective imaging findings as it relates to detection of the primary and metastatic tumors are what is relevant. For patients with a diagnosis of metastatic adenocarcinoma, a PET/CT scan is seldom ordered because it is not part of many standard staging or surveillance regimens for the typical histologic subtypes that metastasize to bone, except for lung carcinoma, for which PET/CT is considered part of the standard staging evaluation. There has historically been no standard protocol for the next steps in evaluating a patient with a skeletal metastasis of unknown primary after the standard diagnostic algorithm. The next steps could include mammography, an endoscopic naso-oropharyngeal examination, upper gastrointestinal endoscopy, colonoscopy, and PET/CT. Therefore, it is very likely that additional patients with skeletal metastasis of unknown primary were not included because their provider never ordered a PET/CT scan or one was never approved by their payer as an additional part of their diagnostic evaluation. This ultimately led to the overall small number of patients in this investigation, which limits the conclusions that can be drawn from the data.

We were surprised to find that PET/CT did not detect the primary neoplasm when the standard diagnostic algorithm failed to do so, identifying only one of 13 unknown primary sites. These findings are in concordance with prior investigations [1, 3, 7, 8]. In one study of primarily soft tissue metastases, PET/CT identified the primary source on 42.5% of occasions [1]. This compares with 66% (23 of 35) of the primary sites in our series, after we included those primary sites that PET/CT confirmed. Similar to a 2014 series [7], our investigation found no difference in PET/CT and standard CT chest/abdomen/pelvis in detecting the primary neoplasm.

Although the PET/CT did not detect additional primary tumors not identified by the standard diagnostic algorithm, it did identify a substantial number of additional metastatic sites. The overall number of detected metastases increased from 115 to 176 after the addition of the PET/CT. Interestingly, only eight of these 61 additional metastases were skeletal metastases that were not detected by whole body bone scan. The performance of PET/CT in the detection of additional metastatic sites was not well evaluated in the prior investigations. Nevertheless, the findings of our study raise the question whether PET/CT, if it is equivalent in detecting the primary compared with the standard diagnostic algorithm, should replace the current standard evaluation protocol. As an additional benefit, the PET/CT could be used to confirm truly oligometastatic disease for the surgeon considering a wide resection of a metastasis for this indication. This question is more complex than it seems on the surface because it could result in the overuse of PET/CT for potentially benign diagnoses (for example, giant cell tumor of bone).


The results of this investigation suggest that PET/CT does not appear to provide additional benefit for the detection of the primary cancer site in patients with a skeletal metastasis from an unknown primary source. Given the observation that it aided the detection of metastatic burden in our cohort, PET/CT may have efficacy in the evaluation of the overall metastatic burden. There were few patients in this single-center study, and we believe that larger, prospective evaluations are needed to confirm or refute the findings of this investigation and determine whether PET/CT has a role in the initial evaluation of skeletal metastases in the setting of a known or unknown primary cancer.


We thank Matthew Rerko, BA, research coordinator for the Cleveland Clinic Taussig Cancer Center, for his assistance with database search and REDCap management.


1. Han A, Xue J, Hu M, Zheng J, Wang X. Clinical value of 18F-FDG PET-CT in detecting primary tumor for patients with carcinoma of unknown primary. Cancer Epidemiol. 2012;36:470-475.
2. Jacobsen S, Stephensen SL, Paaske BP, Lie PG, Lausten GS. Skeletal metastases of unknown origin: a retrospective analysis of 29 cases. Acta Orthop Belg. 1997;63:15-22.
3. Katagiri H, Takahashi M, Inagaki J, Sugiura H, Ito S, Iwata H. Determining the site of the primary cancer in patients with skeletal metastasis of unknown origin: a retrospective study. Cancer. 1999;86:533-537.
4. Piccioli A, Maccauro G, Spinelli MS, Biagini R, Rossi B. Bone metastases of unknown origin: epidemiology and principles of management. J Orthop Traumatol. 2015;16:81-86.
5. Rohren EM, Turkington TG, Coleman RE. Clinical applications of PET in oncology. Radiology. 2004;231:305-332.
6. Rougraff BT, Kneisl JS, Simon MA. Skeletal metastases of unknown origin. A prospective study of a diagnostic strategy. J Bone Joint Surg Am. 1993;75:1276-1281.
7. Shimada H, Setoguchi T, Yokouchi M, Sasaki H, Ishidou Y, Kawamura I, Abematsu M, Nagano S, Komiya S. Metastatic bone tumors: Analysis of factors affecting prognosis and efficacy of CT and (18)F-FDG PET-CT in identifying primary lesions. Mol Clin Oncol. 2014;2:875-881.
8. Takagi T, Katagiri H, Kim Y, Suehara Y, Kubota D, Akaike K, Ishii M, Mukaihara K, Okubo T, Murata H, Takahashi M, Kaneko K, Saito T. Skeletal metastasis of unknown primary origin at the initial visit: a retrospective analysis of 286 cases. PLoS One. 2015;10:e0129428.
9. Ugras N, Yalcinkaya U, Akesen B, Kanat O. Solitary bone metastases of unknown origin. Acta Orthop Belg. 2014;80:139-143.
© 2020 by the Association of Bone and Joint Surgeons