SCIENTIFIC ARTICLES
ASSESSMENT OF VALUE OF ROUTINE BONE SCANS IN PATIENTS WITH NEWLY DIAGNOSED PROSTATE CANCER * GLENN GERBER, M.D. GERALD W. CHODAK, M.D. From the Section of Urology, Department of Surgery, Pritzker School of Medicine, The University of Chicago, Chicago, Illinois
ABSTRACT-The value oj routine bone scans as a staging procedure was assessed in patients with newly diagnosed prostate cancer. Records jrom 277 patients were reviewed retrospectively to determine the serum acid and alkaline phosphatases, the presence or absence oj bone pain, and the results oj bone scans and other radiographic studies at the time oj initial diagnosis. We determined the sensitivity and specijicity oj an abnormal acid phosphatase, an abnormal alkaline phosphatase, and the presence oj bone pain used in combination jor assessing bone metastases. Ij at least one of these three parameters was present, the sensitivity was 97 percent, whereas ij all three tests were normal, the specijicity was 78 percent. The negative predictive value jor all three tests combined is 99 percent. These results suggest that a routine bone scan to stage patients with newly diagnosed prostate cancer who have no bone pain and normal acid and alkaline phosphatases may not be warranted in all cases.
In the U. S., the management of patients with carcinoma of the prostate is based in part on the stage of disease at the time of diagnosis. When the tumor is localized to the prostate, curative surgery or radiation therapy is usually recommended; whereas, if metastases are present, radical prostatectomy or pelvic radiation are rarely performed. The most common sites of spread of prostate cancer are the pelvic lymph nodes and the axial skeleton. At present, the most widely accepted method of detecting skeletal tumor spread is bone scintigraphy. Thus, it is usually performed as a routine staging procedure in all men with newly diagnosed prostate cancer.1,2 While radionuelide imaging is very sensitive in detecting bone metastases, with reported false-negative rates of ·Supported In part by the Whirlpool Foundation and the Cancer and Urology Research Foundation.
418
2 percent for a variety of neoplasms, 3 the specificity is much lower. Increased tracer uptake is seen in benign conditions such as increased osteoid formation or mineralization of osteoid, increased blood flow, or an interrupted sympathetic nerve supply" Thus, inflammation, prior trauma, or metabolic bone disease may yield identical scintigraphic images as those seen with metastatic tumor. Through the use of supplemental procedures, such as plain radiographs, computerized tomography (CT) scans, magnetic resonance images (MRI), or bone biopsy, benign and malignant causes of abnormal scintigrams usually can be distinguished. However, these additional studies may increase the cost, inconvenience, and potential morbidity associated with the detection of bone metastases. Despite its high sensitivity for detecting bone metastases, the value of bone scans as a routine UROLOGY
I
MAY 1991
I
VOLUME XXXVII, NUMBER 5
procedure has never been carefully assessed. At initial presentation, approximately 20 percent to 25 percent of men with prostate cancer have bone metastases. 5 Therefore, the majority of men undergo an unnecessary test. If other tests could be used to select patients who did not need a bone scan, then patient-care would be improved and medical costs would be reduced. Although several parameters, including serum acid and alkaline phosphatases, tumor grade, and skeletal symptoms are useful for predicting bone metastases, none of the factors individually is sufficiently reliable. Recent studies have concluded that routine bone scintigraphy was not warranted for assessing bone metastases in patients with bladder or kidney cancer who had both no bone pain and/or a normal alkaline phosphatase because the false-negative rate was so low.6-8 Prompted by these results, we initiated a study to determine if combining acid and alkaline phosphatases and bone symptoms could identify which patients should have a bone scan as part of their initial evaluation. Material and Methods Records were reviewed retrospectively for all patients with prostate cancer receiving treatment a.t !he University of Chicago Hospitals and ClInICS from 1983-1990. Patients were included even if their original diagnosis of malignancy occurred prior to 1983. Specific parameters recorded from the time of initial diagnosis were serum alkaline phosphatase, serum acid phosphatase (total and prostatic fraction), bone-related symptoms, and the presence of bone metastases. Serum acid and alkaline phosphatase determinations were performed using standard commercially available assays. Serum acid phosphatase was determined by radioimmunoassay in 24 percent and by an enzymatic method in 76 percent of the patients. The normal values were established at each laboratory since some studies were performed at other hospitals. The results of all studies were reviewed for this investigation. The presence of bone metastases was determined by bone scintigraphy in all patients by injecting 25 mCi Tc-99m methylene diphosphonate (MDP) intravenously. Whole body images and pertinent additional views were obtained two hours after tracer injection. All scintigrams were reviewed by nuclear medicine physicians and plain radiographs of areas with increased tracer uptake were obtained when necessary to UROLOGY I MAY 1991 I VOLUME XXXVII, NUMBER 5
Sensitivity and specificity oj parameters in predicting bone metastases Sensitivity (%) Specificity (%) Parameters
TABLE
1.
Acid phosphatase Alkaline phosphatase Bone pain
48157 (84) 38/47 (74) 19/52 (37)
178/208 (86) 148/160 (93) 208/209 (99)
Combined
57/59 (91)
136/175 (78)
establish the presence or absence of bone metastases. If the diagnosis of bone metastases remained unclear after bone scans and plain radiographs were reviewed, additional studies, such as CT, MRI, and/or bone biopsy were obtained at the discretion of the attending urologist after consultation with the radiologist. The presence of bone pain was assessed by reviewing the medical history. If no bone symptoms were stated, then they were recorded as absent. The data were analyzed to determine the sensitivity, specificity, and positive and negative predictive values for patients with either normal prostatic acid and alkaline phosphatases and no bone pain or with at least one of these three criteria abnormal. Results Between 1983 and 1990, 288 patients with prostate cancer were treated at The University) of Chicago. The patients ranged in age frorq forty-one to ninety years (median, 68 years). or", these, 6 had Stage Al disease and therefore: did not undergo a staging workup, 4 patIents refused staging evaluation, and in 1 other the records were unavailable. Thus, 96 percent (277/288) of the patients are included in th!s analysiS. During this same period, 137 additional patients were seen in the outpatient clinic for a second opinion or a consultation, but they were not treated at our institution or they received their initial treatment elsewhere. These patients were not included because the results of the various studies could not be verified. Complete data were available for 74 percent (204/277) of the patients. This included 45/61 (74 %) of the men with metastases and 159/216 (74 %) of the men without metastases. The sensitivity and specificity for detecting bone metastases were determined for abnormal acid phosphatase, abnormal alkaline phosphatase, and the presence of bone symptoms used alone or in combination (Table I). To determine the true positive rate of the combined parameters at least one of the three parameters had to 419
TABLE
II. Predictive value oj parameters in determining bone metastases
Parameter
Predictive Value (%) Positive Negative
Acid phosphatase Alkaline phosphatase Bone pain
48/78 (62) 35/47 (74) 19/20 (95)
178/187 (95)
Combined
57/59 (97)
136/138 (99)
1481160 (93)
208/241 (86)
be abnormal. If one parameter was abnormal but some data were unavailable, patients were still included to assess sensitivity and specificity of the three parameters used in combination because the results for the missing data would not have altered the classification. Of the 61 patients with bone metastases, 57 had at least one parameter abnormal (true positive), and 2 patients with documented bone metastases had all three parameters normal (false negative). The remaining 2 patients with bone metastases had no bone pain and a normal acid phosphatase, but no alkaline phosphatase level was available. Thus, the sensitivity of the combined parameters for patients with complete data is 97 percent (57/59). However, if the alkaline phosphatase had been normal in the 2 patients missing this test, then the sensitivity of these three parameters used in combination for predicting bone metastases would be lowered to 93 percent (57/61). Of the 216 patients without bone metastases, 136 had normal acid and alkaline phosphatases and no bone pain (true negative), and 39 men had at least one of the three parameters abnormal (false positive). Thus, the specificity for all three parameters negative is 78 percent (1361 175). Incomplete data were available on the remaining 41 patients, however, the parameters that were available were all normal. If the missing test results had been abnormal (false positives), then the true specificity would be reduced to 63 percent (136/216). The positive and negative predictive values were also calculated for each individual parameter as well as for the combined parameters (Table II). Similar to the determination of sensitivity and specificity, patients were still included in the assessment of the predictive values if some data were unavailable but at least one of the available parameters was abnormal. The positive predictive value of one or more abnormal parameters was 59 percent (57/96), and the negative predictive value if all three parameters Were normal was 99 percent (1361138). How420
III. Predictive value oj PSA in determining bone metastases Predictive Value (%)
TABLE
PSA Level ~ ~ ~
10 ng/mL 20 ng/mL 25 ng/mL
Positive
Negative
13/48 (27) 13/37 (35) 13/32 (41)
48/48 (100) 43/43 (100) 32/32 (100)
IV. Supplemental studies required to evaluate abnormal bone scan findings
TABLE
Study Plain radiographs CT scan MRI Bone biopsy
- - No. of Studies in Patients With Norm. Parameters Abnorm. Parameters
79 6 3 3
51
o 2 o
ever if we assume that the aforementioned 2 pati~nts with bone metastases and no alkaline phosphatase level available actually had a normal alkaline phosphatase, then the negative predictive value would be reduced to 97 percent (136/140). To further refine the predictive value, We analyzed the results of serum PSA levels which were available in 80 of these patients. Of the 61 men with bone metastases, 13 (21 %) had a PSA level available prior to the onset of treatment of their prostate cancer, as compared with 67 of the 216 (31 %) men without bone metastases. The median PSA values in patients with and without bone metastases were 1,179 ng/mL (range 27.4-9,565 ng/mL) and 12 ng/mL (range 0.2-555 ng/mL), respectively. The predictive value of a PSA level greater than or equal to 10 ng/mL, 20 ng/mL, and 25 ng/mL are shown in Table III. Due to abnormalities observed on bone scintigraphy, 27 percent (581216) of the patients without bone metastases and 34 percent (22/65) of the patients with bone metastases required further study. The supplemental studies needed in patients with and without abnormal parameters are shown in Table IV. Comment Detection of bone metastases in patients with carcinoma of the prostate is of vital importance in selecting appropriate therapy. Because bone scintigraphy is the most sensitive method for. detecting these metastases,3 it is performed UROLOCY I MAY 1991 I VOLUME XXXVII, NUMBER 5
routinely after the diagnosis has been established. Unfortunately, many patients have an abnormal bone scan in the absence of metastases which necessitates additional evaluation. In the present study, 27 percent of the patients without metastases required additional tests, including three open bone biopsies, to rule out skeletal tumor spread. The question we have raised is whether or not greater selectivity could be applied in performing bone scans based on other information obtained as part of the general workup. Because other studies found that serum alkaline phosphatase and bone symptoms could be reliably used to select patients with possible bone metastases from bladder and kidney cancer,6-B we combined these two parameters with serum acid phosphatase to assess if bone metastases could be accurately predicted in patients with prostate cancer. In the bladder and kidney cancer studies, it was noted that 0-2.5 percent of patients with no bone pain andlor a normal alkaline phosphatase had bone metastases. 6- 8 Based on these results it was concluded that bone scans were not indicated in patients with bladder and kidney cancer who had no bone pain andlor a normal alkaline phosphatase level. Similarly, we have found that only 1.4 percent (21138) of the men with normal acid and alkaline phosphatases and no bone pain had evidence of bone metastases (false negative). The false-negative rate could be as high as 2.9 percent (41140) if missing data in 2 patients with bone metastases were normal. This result suggests that a bone scan may no longer be necessary for some patients with newly diagnosed prostate cancer particularly those who are not candidates for radical prostatectomy. Because of the significant implications of this result, we are presently reluctant to recommend omitting bone scintigraphy as part of the routine staging evaluation in all men with newly diagnosed prostate cancer until further studies substantiate our finding. Several shortcomings could have influenced our results. The first of these shortcomings relates to incomplete data in 25 percent of our patients. Unfortunately, in a retrospective review this is unavoidable. However, the finding that only 1.4-2.9 percent of the patients with normal serum phosphatase and no bone pain had bone metastases is not adversely affected by this lack of data. Only 4 of the 61 patients with metastases did not have at least one abnormal parameter. Therefore, even if all the other data did UROLOGY I MAY 1991 I VOLUME XXXVII, NUMBER 5
not support our conclusion, the false-negative rate would be no greater than 2.9 percent (41
140). A second possible criticism of our study involves the patient population selected for analysis. Since all patients were seen at a tertiary care center, it is conceivable that our population is not representative of all men with prostate cancer in the U.S. Based on basic patient information, this does not seem to be the case. The median patient age of sixty-eight years and the frequency of bone metastases at the time of initial diagnosis of prostate cancer (22 %) were similar to those found in the 1982 national survey of prostate cancer conducted by the American College of Surgeons. 5 Nevertheless, referral patterns at a tertiary care center may skew the data leading to unwarranted conclusions. Finally, the reliability of asseSsing bone pain retrospectively by chart review is limited. While acknowledging these shortcomings, we believe that if a prospective study currently in progress supports a false-negative rate of 3 percent in predicting the absence of bone metastases based on symptomatology and laboratory data, it would be reasonable to omit bone scintigraphy as a routine staging test in selected patients with newly diagnosed prostate cancer. Clearly, substantial data would be needed in the group eligible for radical prostatectomy. Further study to substantiate our findings is also warranted because other studies have noted a much higher percentage of patients with bone metastases who had normal acid and alkaline phosphatases. Merrick et al. 9 reported that 22 percent of patients with bone metastases at presentation had normal serum phosphatase levels. An explanation for these disparate findings is not evident. Further refinement in predicting bone metastases in men with prostate cancer may result from the use of prostate-specific antigen (PSA) in conjunction with acid and alkaline phosphatases and bone symptoms. PSA values in men with untreated bone metastases average approximately 500 ng/mL. 1o In comparison, men with localized prostate cancer tend to have PSA levels less than 25 ng/mL. 10 Unfortunately, there is a great deal of overlap of PSA values in both groups. Stamey and Kabalin 10 reported on a group of men with untreated prostate cancer. They noted PSA values as low as 30 ng/mL in men with bone metastases, while men with clinically localized disease had PSA values as high as 230 421
ng/mL in some cases. In our study, no patient with bone metastases had a PSA value less than 25 ng/mL. Because of our limited data, we are unable to draw conclusions regarding the role of PSA in predicting the presence of bone metastases. However, we are continuing to investigate this issue prospectively, and it is conceivable that our false-negative rate of 1.4-2.9 percent could be lowered further through the use of PSA. Our analysis has been performed with the assumption that the bone scan has 100 percent sensitivity. However, a negative bone scan is not absolute proof that metastases are absent. The aforementioned national survey of prostate cancer reported that 4 percent of men with metastatic disease noted on roentgenographic survey had negative bone scans. 5 In addition, one can assume that certain small volumes of tumor metastatic to the skeleton cannot be detected by bone scintigraphy. Confirmation of this would require pathologic studies that are not readily available. However, in one study, in approximately 15 percent of patients undergoing radical prostatectomy for pathologically confirmed local disease metastases later developed. l l While nodal disease could not be excluded as the source of metastases and radiographic surveys rather than bone scans were used in some cases, it is still likely that some recurrent disease was secondary to undetected skeletal tumor spread. Therefore, the objection that unnecessary radical surgery or pelvic irradiation will be performed if bone scans are omitted in some patients is tempered by the realization that this already occurs in a small fraction of cases. Although, ideally, clinicians would like to completely avoid inappropriate therapy in patients with cancer, few tests used for selecting therapy have 100 percent accuracy. As a result of the changing medico-economic climate, physicians may be increasingly forced to choose the best allocation of limited health care resources. One way to accomplish this is to be more selec-
422
tive in obtaining expensive tests when the risk to patients is small. The results from this study suggest that omitting the bone scan in patients with newly diagnosed prostate cancer who have normal acid and alkaline phosphatases and no bone pain may reduce health care costs and unnecessary tests without seriously compromising patient-care. Thus, it seems reasonable to ask if the potential benefit derived from omitting bone scintigraphy in selected patients outweighs the disadvantage of incorrectly staging the very low percentage of additional cases. 5841 South Maryland Avenue Box 403 Chicago, Illinois 60637 (DR. CHODAK) References 1. Paulson DF: Surgical therapy for cancer of the prostate, in Skinner DG, and Leskovsky G (Eds): Diagnosis and Management of Genitourinary Cancer, Philadelphia, WB Saunders Co, 1988, pp 417-424. 2. Catalona WJ, and Scott WN: Carcinoma of the prostate, in Walsh PC, Gittes RF, Perlmutter AD, and Stamey TA (Eds): Campbell's Urology, Philadelphia, WB Saunders Co, 1986, pp 1463-1534. 3. O'Mara RE: Skeletal scanning in neoplastic disease, Cancer 37: 480 (1976). 4. Mettler FA, and Guiberteau MJ: Bone scanning, in Essentials of Nuclear Medicine Imaging, ed 2, Orlando, Florida, Grune and Stratton, Inc., 1986, pp 247-283. 5. Murphy Gp, et all The national survey of prostate cancer in the United States by the American College of Surgeons, J Urol 127: 928 (1982). 6. Lindner A, and DeKernion JB: Cost-effective analysis of pre-cystectomy radioisotope scans, J Uro1128: 1181 (1982). 7. Rosen PR, and Murphy KG: Bone scintigraphy in the initial staging of patients with renal-cell carcinoma: concise communication, J Nucl Med 25: 289 (1984). 8. Berger GL, Sadlowski RW, Sharpe JR, and Finney RP: Lack of value of routine preoperative bone and liver scans in cystectomy candidates, J Uro1125: 637 (1981). 9. Merrick MV, Ding CL, Chisholm GD, and Elton RA: Prognostic significance of alkaline and acid phosphatases and skeletal scintigraphy in carcinoma of the prostate, Br J Urol 57: 715 (1985). 10. Stamey TA, and Kabalin IN: Prostate-specific antigen in the diagnosis and treatment of adenocarcinoma of the prostate. I. Untreated patients, J Uro1141: 1070 (1989). 11. Gibbons RP, Correa RJ Jr, Brannen GE, and Weissman RM: Thtal prostatectomy for clinically localized prostate cancer, long-term result, J Uro1141: 564 (1989).
UROLOGY
I
MAY 1991
I
VOLUME XXXVII, NUMBER 5
ASSESSMENT OF VALUE OF ROUTINE BONE SCANS IN PATIENTS WITH NEWLY DIAGNOSED PROSTATE CANCER * GLENN GERBER, M.D. GERALD W. CHODAK, M.D. From the Section of Urology, Department of Surgery, Pritzker School of Medicine, The University of Chicago, Chicago, Illinois
ABSTRACT-The value oj routine bone scans as a staging procedure was assessed in patients with newly diagnosed prostate cancer. Records jrom 277 patients were reviewed retrospectively to determine the serum acid and alkaline phosphatases, the presence or absence oj bone pain, and the results oj bone scans and other radiographic studies at the time oj initial diagnosis. We determined the sensitivity and specijicity oj an abnormal acid phosphatase, an abnormal alkaline phosphatase, and the presence oj bone pain used in combination jor assessing bone metastases. Ij at least one of these three parameters was present, the sensitivity was 97 percent, whereas ij all three tests were normal, the specijicity was 78 percent. The negative predictive value jor all three tests combined is 99 percent. These results suggest that a routine bone scan to stage patients with newly diagnosed prostate cancer who have no bone pain and normal acid and alkaline phosphatases may not be warranted in all cases.
In the U. S., the management of patients with carcinoma of the prostate is based in part on the stage of disease at the time of diagnosis. When the tumor is localized to the prostate, curative surgery or radiation therapy is usually recommended; whereas, if metastases are present, radical prostatectomy or pelvic radiation are rarely performed. The most common sites of spread of prostate cancer are the pelvic lymph nodes and the axial skeleton. At present, the most widely accepted method of detecting skeletal tumor spread is bone scintigraphy. Thus, it is usually performed as a routine staging procedure in all men with newly diagnosed prostate cancer.1,2 While radionuelide imaging is very sensitive in detecting bone metastases, with reported false-negative rates of ·Supported In part by the Whirlpool Foundation and the Cancer and Urology Research Foundation.
418
2 percent for a variety of neoplasms, 3 the specificity is much lower. Increased tracer uptake is seen in benign conditions such as increased osteoid formation or mineralization of osteoid, increased blood flow, or an interrupted sympathetic nerve supply" Thus, inflammation, prior trauma, or metabolic bone disease may yield identical scintigraphic images as those seen with metastatic tumor. Through the use of supplemental procedures, such as plain radiographs, computerized tomography (CT) scans, magnetic resonance images (MRI), or bone biopsy, benign and malignant causes of abnormal scintigrams usually can be distinguished. However, these additional studies may increase the cost, inconvenience, and potential morbidity associated with the detection of bone metastases. Despite its high sensitivity for detecting bone metastases, the value of bone scans as a routine UROLOGY
I
MAY 1991
I
VOLUME XXXVII, NUMBER 5
procedure has never been carefully assessed. At initial presentation, approximately 20 percent to 25 percent of men with prostate cancer have bone metastases. 5 Therefore, the majority of men undergo an unnecessary test. If other tests could be used to select patients who did not need a bone scan, then patient-care would be improved and medical costs would be reduced. Although several parameters, including serum acid and alkaline phosphatases, tumor grade, and skeletal symptoms are useful for predicting bone metastases, none of the factors individually is sufficiently reliable. Recent studies have concluded that routine bone scintigraphy was not warranted for assessing bone metastases in patients with bladder or kidney cancer who had both no bone pain and/or a normal alkaline phosphatase because the false-negative rate was so low.6-8 Prompted by these results, we initiated a study to determine if combining acid and alkaline phosphatases and bone symptoms could identify which patients should have a bone scan as part of their initial evaluation. Material and Methods Records were reviewed retrospectively for all patients with prostate cancer receiving treatment a.t !he University of Chicago Hospitals and ClInICS from 1983-1990. Patients were included even if their original diagnosis of malignancy occurred prior to 1983. Specific parameters recorded from the time of initial diagnosis were serum alkaline phosphatase, serum acid phosphatase (total and prostatic fraction), bone-related symptoms, and the presence of bone metastases. Serum acid and alkaline phosphatase determinations were performed using standard commercially available assays. Serum acid phosphatase was determined by radioimmunoassay in 24 percent and by an enzymatic method in 76 percent of the patients. The normal values were established at each laboratory since some studies were performed at other hospitals. The results of all studies were reviewed for this investigation. The presence of bone metastases was determined by bone scintigraphy in all patients by injecting 25 mCi Tc-99m methylene diphosphonate (MDP) intravenously. Whole body images and pertinent additional views were obtained two hours after tracer injection. All scintigrams were reviewed by nuclear medicine physicians and plain radiographs of areas with increased tracer uptake were obtained when necessary to UROLOGY I MAY 1991 I VOLUME XXXVII, NUMBER 5
Sensitivity and specificity oj parameters in predicting bone metastases Sensitivity (%) Specificity (%) Parameters
TABLE
1.
Acid phosphatase Alkaline phosphatase Bone pain
48157 (84) 38/47 (74) 19/52 (37)
178/208 (86) 148/160 (93) 208/209 (99)
Combined
57/59 (91)
136/175 (78)
establish the presence or absence of bone metastases. If the diagnosis of bone metastases remained unclear after bone scans and plain radiographs were reviewed, additional studies, such as CT, MRI, and/or bone biopsy were obtained at the discretion of the attending urologist after consultation with the radiologist. The presence of bone pain was assessed by reviewing the medical history. If no bone symptoms were stated, then they were recorded as absent. The data were analyzed to determine the sensitivity, specificity, and positive and negative predictive values for patients with either normal prostatic acid and alkaline phosphatases and no bone pain or with at least one of these three criteria abnormal. Results Between 1983 and 1990, 288 patients with prostate cancer were treated at The University) of Chicago. The patients ranged in age frorq forty-one to ninety years (median, 68 years). or", these, 6 had Stage Al disease and therefore: did not undergo a staging workup, 4 patIents refused staging evaluation, and in 1 other the records were unavailable. Thus, 96 percent (277/288) of the patients are included in th!s analysiS. During this same period, 137 additional patients were seen in the outpatient clinic for a second opinion or a consultation, but they were not treated at our institution or they received their initial treatment elsewhere. These patients were not included because the results of the various studies could not be verified. Complete data were available for 74 percent (204/277) of the patients. This included 45/61 (74 %) of the men with metastases and 159/216 (74 %) of the men without metastases. The sensitivity and specificity for detecting bone metastases were determined for abnormal acid phosphatase, abnormal alkaline phosphatase, and the presence of bone symptoms used alone or in combination (Table I). To determine the true positive rate of the combined parameters at least one of the three parameters had to 419
TABLE
II. Predictive value oj parameters in determining bone metastases
Parameter
Predictive Value (%) Positive Negative
Acid phosphatase Alkaline phosphatase Bone pain
48/78 (62) 35/47 (74) 19/20 (95)
178/187 (95)
Combined
57/59 (97)
136/138 (99)
1481160 (93)
208/241 (86)
be abnormal. If one parameter was abnormal but some data were unavailable, patients were still included to assess sensitivity and specificity of the three parameters used in combination because the results for the missing data would not have altered the classification. Of the 61 patients with bone metastases, 57 had at least one parameter abnormal (true positive), and 2 patients with documented bone metastases had all three parameters normal (false negative). The remaining 2 patients with bone metastases had no bone pain and a normal acid phosphatase, but no alkaline phosphatase level was available. Thus, the sensitivity of the combined parameters for patients with complete data is 97 percent (57/59). However, if the alkaline phosphatase had been normal in the 2 patients missing this test, then the sensitivity of these three parameters used in combination for predicting bone metastases would be lowered to 93 percent (57/61). Of the 216 patients without bone metastases, 136 had normal acid and alkaline phosphatases and no bone pain (true negative), and 39 men had at least one of the three parameters abnormal (false positive). Thus, the specificity for all three parameters negative is 78 percent (1361 175). Incomplete data were available on the remaining 41 patients, however, the parameters that were available were all normal. If the missing test results had been abnormal (false positives), then the true specificity would be reduced to 63 percent (136/216). The positive and negative predictive values were also calculated for each individual parameter as well as for the combined parameters (Table II). Similar to the determination of sensitivity and specificity, patients were still included in the assessment of the predictive values if some data were unavailable but at least one of the available parameters was abnormal. The positive predictive value of one or more abnormal parameters was 59 percent (57/96), and the negative predictive value if all three parameters Were normal was 99 percent (1361138). How420
III. Predictive value oj PSA in determining bone metastases Predictive Value (%)
TABLE
PSA Level ~ ~ ~
10 ng/mL 20 ng/mL 25 ng/mL
Positive
Negative
13/48 (27) 13/37 (35) 13/32 (41)
48/48 (100) 43/43 (100) 32/32 (100)
IV. Supplemental studies required to evaluate abnormal bone scan findings
TABLE
Study Plain radiographs CT scan MRI Bone biopsy
- - No. of Studies in Patients With Norm. Parameters Abnorm. Parameters
79 6 3 3
51
o 2 o
ever if we assume that the aforementioned 2 pati~nts with bone metastases and no alkaline phosphatase level available actually had a normal alkaline phosphatase, then the negative predictive value would be reduced to 97 percent (136/140). To further refine the predictive value, We analyzed the results of serum PSA levels which were available in 80 of these patients. Of the 61 men with bone metastases, 13 (21 %) had a PSA level available prior to the onset of treatment of their prostate cancer, as compared with 67 of the 216 (31 %) men without bone metastases. The median PSA values in patients with and without bone metastases were 1,179 ng/mL (range 27.4-9,565 ng/mL) and 12 ng/mL (range 0.2-555 ng/mL), respectively. The predictive value of a PSA level greater than or equal to 10 ng/mL, 20 ng/mL, and 25 ng/mL are shown in Table III. Due to abnormalities observed on bone scintigraphy, 27 percent (581216) of the patients without bone metastases and 34 percent (22/65) of the patients with bone metastases required further study. The supplemental studies needed in patients with and without abnormal parameters are shown in Table IV. Comment Detection of bone metastases in patients with carcinoma of the prostate is of vital importance in selecting appropriate therapy. Because bone scintigraphy is the most sensitive method for. detecting these metastases,3 it is performed UROLOCY I MAY 1991 I VOLUME XXXVII, NUMBER 5
routinely after the diagnosis has been established. Unfortunately, many patients have an abnormal bone scan in the absence of metastases which necessitates additional evaluation. In the present study, 27 percent of the patients without metastases required additional tests, including three open bone biopsies, to rule out skeletal tumor spread. The question we have raised is whether or not greater selectivity could be applied in performing bone scans based on other information obtained as part of the general workup. Because other studies found that serum alkaline phosphatase and bone symptoms could be reliably used to select patients with possible bone metastases from bladder and kidney cancer,6-B we combined these two parameters with serum acid phosphatase to assess if bone metastases could be accurately predicted in patients with prostate cancer. In the bladder and kidney cancer studies, it was noted that 0-2.5 percent of patients with no bone pain andlor a normal alkaline phosphatase had bone metastases. 6- 8 Based on these results it was concluded that bone scans were not indicated in patients with bladder and kidney cancer who had no bone pain andlor a normal alkaline phosphatase level. Similarly, we have found that only 1.4 percent (21138) of the men with normal acid and alkaline phosphatases and no bone pain had evidence of bone metastases (false negative). The false-negative rate could be as high as 2.9 percent (41140) if missing data in 2 patients with bone metastases were normal. This result suggests that a bone scan may no longer be necessary for some patients with newly diagnosed prostate cancer particularly those who are not candidates for radical prostatectomy. Because of the significant implications of this result, we are presently reluctant to recommend omitting bone scintigraphy as part of the routine staging evaluation in all men with newly diagnosed prostate cancer until further studies substantiate our finding. Several shortcomings could have influenced our results. The first of these shortcomings relates to incomplete data in 25 percent of our patients. Unfortunately, in a retrospective review this is unavoidable. However, the finding that only 1.4-2.9 percent of the patients with normal serum phosphatase and no bone pain had bone metastases is not adversely affected by this lack of data. Only 4 of the 61 patients with metastases did not have at least one abnormal parameter. Therefore, even if all the other data did UROLOGY I MAY 1991 I VOLUME XXXVII, NUMBER 5
not support our conclusion, the false-negative rate would be no greater than 2.9 percent (41
140). A second possible criticism of our study involves the patient population selected for analysis. Since all patients were seen at a tertiary care center, it is conceivable that our population is not representative of all men with prostate cancer in the U.S. Based on basic patient information, this does not seem to be the case. The median patient age of sixty-eight years and the frequency of bone metastases at the time of initial diagnosis of prostate cancer (22 %) were similar to those found in the 1982 national survey of prostate cancer conducted by the American College of Surgeons. 5 Nevertheless, referral patterns at a tertiary care center may skew the data leading to unwarranted conclusions. Finally, the reliability of asseSsing bone pain retrospectively by chart review is limited. While acknowledging these shortcomings, we believe that if a prospective study currently in progress supports a false-negative rate of 3 percent in predicting the absence of bone metastases based on symptomatology and laboratory data, it would be reasonable to omit bone scintigraphy as a routine staging test in selected patients with newly diagnosed prostate cancer. Clearly, substantial data would be needed in the group eligible for radical prostatectomy. Further study to substantiate our findings is also warranted because other studies have noted a much higher percentage of patients with bone metastases who had normal acid and alkaline phosphatases. Merrick et al. 9 reported that 22 percent of patients with bone metastases at presentation had normal serum phosphatase levels. An explanation for these disparate findings is not evident. Further refinement in predicting bone metastases in men with prostate cancer may result from the use of prostate-specific antigen (PSA) in conjunction with acid and alkaline phosphatases and bone symptoms. PSA values in men with untreated bone metastases average approximately 500 ng/mL. 1o In comparison, men with localized prostate cancer tend to have PSA levels less than 25 ng/mL. 10 Unfortunately, there is a great deal of overlap of PSA values in both groups. Stamey and Kabalin 10 reported on a group of men with untreated prostate cancer. They noted PSA values as low as 30 ng/mL in men with bone metastases, while men with clinically localized disease had PSA values as high as 230 421
ng/mL in some cases. In our study, no patient with bone metastases had a PSA value less than 25 ng/mL. Because of our limited data, we are unable to draw conclusions regarding the role of PSA in predicting the presence of bone metastases. However, we are continuing to investigate this issue prospectively, and it is conceivable that our false-negative rate of 1.4-2.9 percent could be lowered further through the use of PSA. Our analysis has been performed with the assumption that the bone scan has 100 percent sensitivity. However, a negative bone scan is not absolute proof that metastases are absent. The aforementioned national survey of prostate cancer reported that 4 percent of men with metastatic disease noted on roentgenographic survey had negative bone scans. 5 In addition, one can assume that certain small volumes of tumor metastatic to the skeleton cannot be detected by bone scintigraphy. Confirmation of this would require pathologic studies that are not readily available. However, in one study, in approximately 15 percent of patients undergoing radical prostatectomy for pathologically confirmed local disease metastases later developed. l l While nodal disease could not be excluded as the source of metastases and radiographic surveys rather than bone scans were used in some cases, it is still likely that some recurrent disease was secondary to undetected skeletal tumor spread. Therefore, the objection that unnecessary radical surgery or pelvic irradiation will be performed if bone scans are omitted in some patients is tempered by the realization that this already occurs in a small fraction of cases. Although, ideally, clinicians would like to completely avoid inappropriate therapy in patients with cancer, few tests used for selecting therapy have 100 percent accuracy. As a result of the changing medico-economic climate, physicians may be increasingly forced to choose the best allocation of limited health care resources. One way to accomplish this is to be more selec-
422
tive in obtaining expensive tests when the risk to patients is small. The results from this study suggest that omitting the bone scan in patients with newly diagnosed prostate cancer who have normal acid and alkaline phosphatases and no bone pain may reduce health care costs and unnecessary tests without seriously compromising patient-care. Thus, it seems reasonable to ask if the potential benefit derived from omitting bone scintigraphy in selected patients outweighs the disadvantage of incorrectly staging the very low percentage of additional cases. 5841 South Maryland Avenue Box 403 Chicago, Illinois 60637 (DR. CHODAK) References 1. Paulson DF: Surgical therapy for cancer of the prostate, in Skinner DG, and Leskovsky G (Eds): Diagnosis and Management of Genitourinary Cancer, Philadelphia, WB Saunders Co, 1988, pp 417-424. 2. Catalona WJ, and Scott WN: Carcinoma of the prostate, in Walsh PC, Gittes RF, Perlmutter AD, and Stamey TA (Eds): Campbell's Urology, Philadelphia, WB Saunders Co, 1986, pp 1463-1534. 3. O'Mara RE: Skeletal scanning in neoplastic disease, Cancer 37: 480 (1976). 4. Mettler FA, and Guiberteau MJ: Bone scanning, in Essentials of Nuclear Medicine Imaging, ed 2, Orlando, Florida, Grune and Stratton, Inc., 1986, pp 247-283. 5. Murphy Gp, et all The national survey of prostate cancer in the United States by the American College of Surgeons, J Urol 127: 928 (1982). 6. Lindner A, and DeKernion JB: Cost-effective analysis of pre-cystectomy radioisotope scans, J Uro1128: 1181 (1982). 7. Rosen PR, and Murphy KG: Bone scintigraphy in the initial staging of patients with renal-cell carcinoma: concise communication, J Nucl Med 25: 289 (1984). 8. Berger GL, Sadlowski RW, Sharpe JR, and Finney RP: Lack of value of routine preoperative bone and liver scans in cystectomy candidates, J Uro1125: 637 (1981). 9. Merrick MV, Ding CL, Chisholm GD, and Elton RA: Prognostic significance of alkaline and acid phosphatases and skeletal scintigraphy in carcinoma of the prostate, Br J Urol 57: 715 (1985). 10. Stamey TA, and Kabalin IN: Prostate-specific antigen in the diagnosis and treatment of adenocarcinoma of the prostate. I. Untreated patients, J Uro1141: 1070 (1989). 11. Gibbons RP, Correa RJ Jr, Brannen GE, and Weissman RM: Thtal prostatectomy for clinically localized prostate cancer, long-term result, J Uro1141: 564 (1989).
UROLOGY
I
MAY 1991
I
VOLUME XXXVII, NUMBER 5
