Noninvasive imaging with thallium-201 scintigraphy may not correlate with survival in patients with osteosarcoma




Histologic response to preoperative chemotherapy is a strong prognostic factor for osteosarcoma (OS). Thallium-201 (Tl-201) scintigraphic response to initial chemotherapy has previously been described as a predictor of histologic response. In the current study, the authors re-examined a series of patients studied using Tl-201 scintigraphy to determine the correlation between changes observed on Tl-201 scintigraphy and event-free survival (EFS).


A total of 22 patients with biopsy-proven OS of the extremity underwent Tl-201 imaging before and immediately after preoperative chemotherapy. The maximum pixel counts taken over the tumor divided by those taken of a background region yielded a tumor-to-background ratio (TBR). The percentage of change in the TBR before and after adjuvant chemotherapy, defined as the alteration ratio (AR), was correlated with EFS.


The median AR was 85% (range, 28-100%). The 3-year EFS was 0.72 (95% confidence interval [95% CI], 0.48-0.86) and the 5-year EFS was 0.67 (95% CI, 0.43-0.86). There was no association between AR and EFS detected in this cohort (hazard ratio, 0.99; 95% CI, 0.95-1.02 [Somers rank correlation coefficient, 0.15]).


Although Tl-201 scintigraphy was used as a tool for the assessment of response to chemotherapy in patients with OS, the AR did not appear to be predictive of EFS in this small group of patients. It is necessary to use the outcome variables of ultimate interest–EFS and overall survival– and not rely on surrogates for outcome to evaluate potential prognostic factors. Cancer, 2010. © 2010 American Cancer Society.

Osteosarcoma (OS) is the most commonly occurring primary malignant bone tumor in children, adolescents, and young adults, with an incidence of approximately 600 cases each year in the United States.1 The successful treatment of OS requires multiagent chemotherapy,2, 3 as well as definitive surgical resection of the primary site of disease and any clinically detectable sites of metastatic disease. Although combination chemotherapy has significantly improved outcomes for patients with OS, risk-adapted therapy has not come of age because of a lack of adequately robust prognostic factors. Currently, prognostic factors for patients with OS include site and size of the primary tumor, the presence of detectable metastatic disease,4-6 adequacy of surgical resection, and degree of tumor necrosis after induction chemotherapy.7, 8 A Huvos grade III or IV response (≥90% necrosis) to induction chemotherapy is reported to be predictive of an improved event-free survival (EFS).9, 10 Tumor necrosis can be assessed only once and only after a relatively protracted period of treatment. For this reason, noninvasive methods have been sought to serve as surrogate markers of this response. The hope has been that these markers could predict EFS and potentially serve as early indicators of response to therapy, which could allow for therapy modifications in poor responders without waiting for the completion of induction chemotherapy.

One of the early functional imaging modalities evaluated as a predictor of response to chemotherapy in patients with OS was thallium-201 (Tl-201) scintigraphy. Tl-201 is a monovalent cationic radioisotope that has properties of a potassium analog.11 Although Tl-201 scintigraphy was initially used for myocardial perfusion imaging, in the 1970s its use in tumor imaging was recognized.12 It has been suggested that patients with OS who did not have a >80% reduction in Tl-201 uptake after preoperative chemotherapy be considered candidates for aggressive surgical treatment.13

In the current study, we describe a follow-up to a previously published report from our institution that demonstrated an excellent correlation between changes on Tl-201 scintigraphy and percentage of tumor necrosis after induction chemotherapy.14 We extended those observations by assessing the correlation between changes observed on Tl-201 scintigraphy and EFS.


Patient Population

A retrospective chart review was conducted to extend the follow-up of a series of patients previously described by Imbriaco et al in 1997.14 This series included 22 consecutive patients (9 females and 13 males with a mean age of 14.3 ± 5 years) with biopsy-proven, high-grade skeletal OS diagnosed at Memorial Sloan-Kettering Cancer Center (MSKCC) between October 1993 and October 1996. Initial staging evaluation included Tl-201 scintigraphy in all patients. All patients received induction chemotherapy before undergoing surgical resection. The majority of patients received chemotherapy on or according to the Children's Cancer Group CCG7921 protocol with cisplatin, doxorubicin, and high-dose methotrexate. One patient was treated on the MSKCC T12 protocol with high-dose methotrexate, vincristine, doxorubicin, cisplatin, cyclophosphamide, and bleomycin and 1 patient was treated with ifosfamide, doxorubicin, and high-dose methotrexate. After the completion of preoperative chemotherapy and before definitive tumor resection, Tl-201 imaging was repeated. No additional chemotherapy was given between the second Tl-201 imaging and surgical resection. The initial histologic diagnosis was confirmed after surgical resection. Tumor response to chemotherapy was graded according to the criteria of Huvos et al.15 The institutional review board at MSKCC approved the review of medical records for this analysis.

Thallium-201 Scintigraphy

The acquisition of images and determination of an alteration ratio (AR) in each patient were described previously.14 Briefly, the maximum pixel counts taken over the tumor divided by the maximum pixel counts taken over a background region yielded a tumor-to-background ratio (TBR). The TBR obtained before chemotherapy was compared with the TBR obtained after chemotherapy, and the percentage change in the TBR was defined as the AR.

Statistical Analysis

The goal of this follow-up study was to determine the association between a Tl-201 scintigraphy-generated AR and EFS. EFS was calculated from the date of surgical resection to date of death, disease recurrence, or last follow-up. The Kaplan-Meier method was used to estimate the EFS function. The Somers rank correlation coefficient was estimated, taking into account that EFS was subject to censoring.16 A univariate Cox regression model was used to examine the association between AR and EFS. Hazard ratios (HRs) and 95% confidence intervals (95% CIs) were reported. All statistical analyses were conducted using SAS 9.1 (SAS Institute Inc, Cary, NC) and R 2.9.2 (SAS Institute Inc) statistical software.


All patients completed pre- and postchemotherapy Tl-201 scans, induction chemotherapy, and surgical resection of the tumor. Patient characteristics, scintigraphic results, and histologic response to chemotherapy are presented in Table 1. In the prior article, the correlation between AR and necrosis was described for 24 patients.14 Review of patient records for the current analysis revealed that 2 of the subjects in the previous analysis did not have OS. Therefore, we excluded them and used only the 22 patients with confirmed OS for the current analysis.

Table 1. Patient Characteristics (n = 22)
  1. CCG indicates Children's Cancer Group; MSKCC, Memorial Sloan-Kettering Cancer Center; Ifos, ifosfamide; Doxo, doxorubicin; Mtx, methotrexate.

Age, y 
Primary tumor location 
 Distal femur14
 Proximal tibia5
 Proximal humerus3
Metastases at diagnosis 
Initial chemotherapy 
Surgical approach 
 Limb salvage20
 Wide amputation2
Huvos grade 
 I (<50% necrosis)3
 II (50-89% necrosis)3
 III (≥90% necrosis)9
 IV (100% necrosis)7
Alteration ratio, % 

The median AR was 85% (range, 28-100%). Three patients had tumors with a Huvos grade I response to chemotherapy, 3 had a grade II response, 9 had a grade III response, and 7 had a grade IV response. The initial publication generated from this series of patients described a strong positive correlation (correlation coefficient, 0.93; standard error, 10.2% [P < .001]) between the AR and the percentage of tumor necrosis at the time of definitive surgical resection.14

All patients resumed chemotherapy postoperatively on or according to their original chemotherapy protocols. No changes were made to chemotherapy regimens based on histologic response to initial chemotherapy. Three patients did not complete planned adjuvant chemotherapy: 2 patients died of disease after the first cycle of postoperative chemotherapy and 1 patient who developed disease progression at a site of metastatic disease after 2 cycles of postoperative chemotherapy was switched from the CCG7921A protocol to a regimen of ifosfamide and doxorubicin.

Six of the 22 patients experienced disease recurrence (3 patients in the lungs, 1 patient in bone, and 2 patients in both the lungs and bone). All patients with recurrent disease had died by the time of last follow-up, although 1 of these patients died of sepsis while in complete response (CR) after the successful treatment of disease recurrence. One patient with metastatic disease at the time of diagnosis never achieved a CR and died of disease 5 months after diagnosis. The remaining 15 patients were alive without evidence of disease at the time of last follow-up. The median follow-up time for survivors without disease progression was 145 months (range, 28-174 months). The 3-year EFS for all patients was 0.72 (95% CI, 0.48-0.86) and the 5-year EFS was 0.67 (95% CI, 0.43-0.86). There was no association noted in this cohort between AR and EFS (HR, 0.99; 95% CI, 0.95-1.02 [Somers rank correlation coefficient, 0.15]). When the 4 patients with metastatic disease were excluded, the 3-year EFS was 0.83 (95% CI, 0.5-0.94). Figure 1 describes EFS in all patients and in the patients with localized disease whose Tl-201 scintigraphic response to chemotherapy yielded an AR greater than or equal to the median AR of 85% and those whose response yielded a value less than the median AR of 85%. Because of the small number of events (n = 3) in the patient group with localized disease, the association between AR and EFS could not be tested using univariate analysis.

Figure 1.

Event-free survival by thallium-201 scintigraphic response (alteration ratio [AR]) to preoperative chemotherapy is shown for (A) all patients (n = 22) and (B) patients with localized disease (n = 18).


It is well established that histologic response to preoperative chemotherapy is correlated with EFS in patients with OS.7-9, 17 A previous report from our institution identified the percentage change in uptake of T1-201 before and after preoperative chemotherapy as being strongly correlated with the percentage of tumor necrosis documented at the time of tumor resection.14 Multiple other groups similarly reported thallium imaging as an indicator of response to the preoperative treatment of OS.18-23 Small studies have suggested that T1-201 scintigraphy correlates more significantly with histologic response to treatment than either digital subtraction angiography19 or technetium-99m methylene-diphosphonate (MDP) bone scintigraphy.22 It is therefore tempting to extrapolate Tl-201 scintigraphic response as a predictor of outcome in patients with OS. The current study expands on our previous institutional report to assess the association between Tl-201 AR and EFS.

In contrast to the positive correlation demonstrated between AR and histologic response to preoperative chemotherapy, no association was demonstrated between AR and EFS. This study population is representative of the general OS population. Four of the 22 patients had metastatic disease at the time of diagnosis, the majority of cases involved the distal femur or proximal tibia, and we observed a 5-year EFS rate of 67%. Because the series of patients was diagnosed and initially treated >13 years ago, the results are unlikely to change with further follow-up. However, the current study is limited by the small sample size and its retrospective nature. Given the limited number of events, a correlation between AR and EFS could exist, and we cannot rule out the possibility of a false-negative result. In addition, patients with both localized and metastatic disease were included in the analysis of an association between AR and EFS. When patients with localized disease were examined independently, the analysis was further limited by the small number of events in this group, and no reliable conclusions could be drawn. If Tl-201 scintigraphy had not been replaced by dynamic enhanced magnetic resonance imaging (MRI) and [F-18]-fluorodeoxy-D-glucose positron emission tomography (18FDG-PET) as the imaging modalities of interest for the assessment of early response to chemotherapy, confirmation of the lack of an association between AR and EFS in a larger prospective study would be required.

Several groups have reported on the use of dynamic enhanced MRI24;25 and 18FDG-PET26-28 to evaluate response to induction chemotherapy in patients with OS. To the best of our knowledge, the majority of reports published to date have compared the results of noninvasive imaging with the degree of necrosis after initial chemotherapy. A correlation between imaging and necrosis cannot be equated with a correlation between imaging and survival. Each surrogate marker must be validated independently as in the work of Hawkins et al.28 In 2002, Hawkins et al reported that both the standard uptake value (SUV) after induction chemotherapy and the change in SUV with induction chemotherapy were correlated with the histopathologic assessment of response in pediatric patients with OS and the Ewing sarcoma family of tumors.28 They recently expanded on this series of patients and published their experience with 18FDG-PET as a predictor of outcome in patients with extremity OS. They reported an association between an SUV <2.5 after induction chemotherapy and improved progression-free survival, which did not achieve a conventional level of significance for patients with localized disease, and they suggested a larger prospective study before establishing 18FDG-PET response as a prognostic factor for patients with OS.29

Tl-201 scintigraphy is no longer an imaging tool that is used commonly in the monitoring of patients with OS; however, the conclusions drawn from its use in this group of patients highlight the caution that should be taken in assessing prognostic factors. Although functional imaging modalities are being evaluated as early indicators of response in many cancers, it is critical to use the outcome variables of ultimate interest–EFS and overall survival– and not settle for surrogates for outcome.


The authors made no disclosures.