Ink J. Rndiarm Onco/o~y RIOI Phys., Vol. Printed in the U.S.A. All rights reserved.
24, pp.
0360.3016/92 $5.00 + .oO Copyright 8 1992 Pergamon Press Ltd.
431434
??Clinical Original Contribution
NUCLEAR ROUNDNESS FACTOR AND LOCAL FAILURE FROM DEFINITIVE RADIATION THERAPY FOR PROSTATIC CARCINOMA JAMES SHAEFFER, CHRISTINE
PH.D.,
B. PHILPUT,
JAMES A. TEGELER, PH.D.
AND ANAS
M.D.,
DEBORAH
M. EL-MAHDI,
A. KUBAN,
M.D.,
Sc.D.,
M.D., FACR
Department of Radiation Oncology and Biophysics, Eastern Virginia Medical School, Norfolk, VA Of 375 patients with prostatic carcinoma treated definitively with radiation therapy at this institution with at least a 5 year follow-up, 23 patients failed locally only, 72 failed with distant metastasis only, 60 had both local and distant failure, while 220 showed no evidence of disease. In search for a possible marker for local failure following radiation therapy, we examined several nuclear morphometric parameters which have been shown to correlate with the biologic aggressiveness of this disease. The 23 locally failed only patients were matched with 23 no evidence of disease patients for stage, grade, treatment modality, prior surgery, age at diagnosis and race. Archival hematoxylin and eosin slides were obtained for 22 of the 23 matched pairs, and morphometric features, including nuclear roundness factor and nuclear area, as well as numbers of nucleoli were assessed using computer-assisted image analysis in both tumor cells and normal prostatic epithelium. Tumor nuclei from the locally failed only patients had significantly higher nuclear roundness factor values (p = 0.0089) compared with tumor cells from no evidence
of disease patients. Analysis of these data by clinical stage demonstrated no significant differences between the locally failed only and no evidence of disease patients. Likewise, there were no significant differences in nuclear roundness factor values of locally failed only and no evidence of disease patients with poorly or moderately welldifferentiated tumors. However, there was a highly significant difference (p = 0.0012) in the nuclear roundness factor values of locally failed only and no evidence of disease patients with well-differentiated tumors. Thus, there appears to be a subset of patients with well-differentiated adenocarcinoma of the prostate who have significantly more irregular tumor nuclei and who fail locally only following definitive radiation therapy. Prostate carcinoma, Radiotherapy,
Nuclear roundness factor, Local failure.
categorize the variable biological behavior of prostate cancer. Nuclear morphology (5, 7, 18, 20, 2 I), motility ( 19) DNA content and various indices of cellular proliferation ( 1,2,8, 11, l&23), growth factor (I 2) oncogenes
INTRODUCTION Prostatic cancer occurs frequently and the incidence will most likely continue to rise in the 1990’s and beyond because of the “aging” of the U.S. male population
(10, 24, 25) and suppressor genes (4) have been investigated for their ability to serve as markers of patient outcome for prostatic carcinoma. We report here on the correlation of nuclear morphology with clinical outcome in a retrospective study involving patients with prostatic carcinoma given definitive radiation therapy.
and ever-increasing emphasis on screening and early detection ( 17). There is frequently great variation in the biologic behavior of prostatic tumor cells, even when they appear histologically similar. Consequently, the clinical course can be unpredictable. Some patients will have widespread disease within a relatively short time after diagnosis whereas others will have a relatively indolent disease which does not significantly alter life span, and which requires little or no clinical intervention. A major challenge will be to search for and identify prognostic indicators or markers which more accurately
METHODS
AND
MATERIALS
In a series of 375 patients with prostatic carcinoma treated definitively at this institution with radiation therapy and having a minimum of 5 years’ follow-up, the
Presented at the 32nd Annual ASTRO Meeting, Miami Beach, FL, 15- 19 October 1990. Reprint requests to: James Shaeffer, Ph.D., 600 Gresham Dr., Norfolk, VA 23507. Acknowledgements-The authors wish to acknowledge Karen H. Manias for her excellent technical assistance, Therese Babb and Claudia G. Brigham for their efforts in obtaining clinical
information, and Stephanie M. White and Bernard Weintraub for their help in preparing the manuscript. This investigation was supported in part by a grant from the Sentara Cancer Center. Dr. El-Mahdi is a recipient of The American Cancer Society Professorship of Clinical Oncology. Accepted for publication 17 March 1992. 431
432
1. J. Radiation Oncology 0 Biology 0 Physics
clinical results were as follows: 220 (59%) had no evidence of disease (NED); 72 (19%) had distant metastasis only (MET); 60 (16%) had distant metastasis and local failure (MET + LF); 23 (6%) had local failure only (LFO). When analyzed by histologic grade and clinical stage, the results were as shown in Table 1. As in previous studies (6, 15) the rate of local recurrence increased with more advanced histologic grade and clinical stage of the tumor. The total rate of local recurrence may be determined by combining the local failure with metastasis groups (MET + LF) with the local failure only groups (LFO). Poorly differentiated (PD) tumors had significantly greater (p < 0.00 1 by chi-square) rates of total local failure compared with well-differentiated (WD) tumors. Patients with stage AZ disease had significantly lower rates (p < 0.001 by chi-square) of total local failure compared with patients with Stage C disease. Likewise, the rate of distant failure also increased with increasing grade and stage. The total rate of distant failure is the combination of metastasis only groups (MET) with the local failure with distant metastasis groups (MET + LF). Poorly differentiated tumors again had significantly higher rates (p < 0.001 by Chi-square) compared to WD tumors, and both A2 and B, patients had lower rates of distant failure (p < 0.00 1 by Chi-square) compared with Stage C patients. Nuclear morphometric measures have been correlated with the biologic behavior of prostatic carcinoma (5, 7, 18, 20, 21). Likewise, nuclear area (3) and nucleolar prominence (3, 13, 14) have been widely used as diagnostic criteria for prostatic carcinoma. In search of a possible marker for local failure from radiation therapy, we compared nuclear morphometric measures and nucleoli in LFO patients versus a matched set of NED patients. The selection of these two groups eliminated possible confounding factors from distant metastasis. The 23 LFO patients were matched with 23 NED patients on the basis of previously reported prognostic factors (9, 22): ??
Histologic grade: well differentiated (WD), moderately differentiated (MD), or poorly differentiated (PD); Table 1. Metastasis and local failure by grade and stage MET only
MET + LF
LFO
Volume 24, Number 3, 1992
Fig. 1. Illustrative examples of shape and nuclear roundness factor. A perfect circle has a nuclear roundness factor (NRF) value of 1.00. Greater degrees of eccentricity with a smooth surface will increase NRF as shown by nuclei with NRF values of 1.06 and 1.37. Also, roughness of the edge will also contribute to increased NRF values, such as the nuclei with values of I. 18 and 1.24. (Modified from Diamond et al. (5) with permission.)
Clinical stage: B1, B2 or C; ?? Prior surgery, for example, transurethral the prostate (TURP); ?? Age at diagnosis; ?? Race: black, white, other. ??
Patients were also matched by treatment modality, either external beam or interstitial I- 125. Archival hematoxylin and eosin slides were obtained for 22 of the 23 matched pairs. Appropriate areas containing tumor and normal prostatic epithelium were reviewed by the same pathologist (J.A.T.), given Gleason (9) scores as the sum of the primary plus secondary scores, and then “blind” coded. Slides were viewed under a 100X oil immersion lens. Nuclear roundness factor (NRF), nuclear area, and numbers of nucleoli per cell were assessed Table 2. Characteristics
Characteristic Treatment modality External beam I- 125 implant Clinical stage BI J-52
Histologic grade WD MD PD Clinical stage A2
Bl B2 C
No. (%)
No. (W)
No. (%)
1 l/l38 (7.9) 31/134 (23.1) 30/103 (29.1)
IO/l38 (7.2) 22/134 (16.4) 28/103 (27.2)
7/138 (5.1) 10/134 (7.5) 6/103 (5.8)
5/49 3/27 24/146 40/153
2/49 0127 23/146 35/153
O/49 3/27 9/146 1 l/153
(10.2) (11.1) (16.4) (26.1)
(4.1) (0) (15.8) (22.9)
(0) (11.1) (6.2) (7.2)
MET = Distant metastasis; MET + LF = Distant metastasis and local failure; LFO = Local failure only.
resection of
C Histologic grade WD MD PD Prior TURP Yes No Age at diagnosis Race Black White
of the study population LF no. patients
NED no. patients
18 4
18 4
3 8 11
3 8 11
7 10 5
7 11 4
8 14 66.2
8 14 68.4
5 17
2 20
LF = Local failure; NED = No evidence
of disease.
433
NRF and prostatic carcinoma 0 J. SHAEFFERet al. Table 3. Summary
Nuclear
vs. normal area, p* area, CL* vs. normal cell cell vs. normal
morphometric
1.082 + 0.027 1.040 + 0.018 0.000 1 34.18 k 11.69 26.28 + 4.82 0.0078 0.71 f 0.21 0.33 + 0.18 0.000 1
1.062 f 0.021 1.037 + 0.009 0.000 1 35.66 + 10.69 29.90 f 7.14 NS 0.82 + 0.13 0.56 t 0.23 0.0008
of 50 tumor nuclei and 50 normal nuclei using the Zeiss Videoplan image analysis system. Areas of tumor and normal prostatic epithelium were selected by the pathologist (J.A.T.) with care to avoid areas ofedge artifact, crush artifact and cautery artifact. Nuclear roundness factor is defined as (nuclear perimeter/27r) + (nuclear area/?r)0.5. Nuclear roundness factor values for a perfect circle are 1.Oand increase above 1.O as the shape of the nucleus becomes more ellipsoid or if the nuclear membrane is more irregular, as illustrated in Figure 1. on a minimum
RESULTS The clinical characteristics of the local failure and NED patients are presented in Table 2. A summary of the results of nuclear parameters is given in Table 3. In each group (LFO and NED), tumor nuclei were shaped more irregularly (had higher values of NRF), were larger, and had more nucleoli compared with normal prostatic epithelial nuclei. Comparing LFO to NED populations, however, tumor NRF values were elevated in the LFO group, while the NED group had more nucleoli in the normal nuclei. When tumor NRF was further studied by treatment modality, grade, and stage (Table 4), the difference in NRF values between the two groups was significant only in those with well-differentiated tumors. Similarly, a significant nuclear roundness
p-value * LFO vs. NED
NED
* p-values determined by Student’s t-test with p = 0.0 1 as level of significance: disease; NRF = nuclear roundness factor.
Table 4. Tumor
data
LFO
parameter
Tumor NRF Normal NRF p-value, tumor Tumor nuclear Normal nuclear p-value, tumor Nucleoli/tumor Nucleoli/normal p-value, tumor
of nuclear
0.0070 NS NS NS NS 0.0039
LFO = Local failure only; NED = No evidence
difference in nucleoli/normal cell nucleus was demonstrated only in those with Stage C disease (Table 5). DISCUSSION Relative NRF is the absolute NRF value of the tumor cells divided by the absolute NRF of the normal prostatic epithelial nuclei. Relative NRF values may be preferable to absolute NRF values because relative NRF values may minimize various fixation artifacts (20, 5). We used absolute NRF values in our study, however, because suitable fields of normal prostatic epithelium were found in only 3 1 of the 44 cases analyzed. As might be expected (Table 3) tumor cells from both local failure and NED groups had greater numbers of nucleoli/cell than normal cells. This may reflect the greater degree of aneuploidy presumed in the malignant cells compared with normal (diploid) cells. We have no explanation for the observation of significantly greater numbers of nucleoli/cell in the normal cells of NED patients compared with LFO patients. Since differences between nucleoli/cell in NED and LFO patients were seen only in normal, but not tumor cells, a more detailed analysis of the differences based on grade and stage were presented in Table 5. A corresponding table for tumor cells was not presented since the differences were not significant. Tumors which failed locally had higher absolute NRF
factor data
Table 5. Nucleolus/normal Group
LFO
All patients External beam I- 125 implant RI
1.082 + 0.027 1.08 1 t 0.027 1.086 r 0.028 1.080 f 0.025 1.08 1 ‘- 0.024 1.083 t 0.031 1.085 k 0.015 1.074 f 0.03 1 1.095 +- 0.029
B2
C WD MD PD
NED 1.062 1.066 1.043 1.047 1.062 1.066 1.051 1.063 1.078
of
+ + + f f f + f f
0.02 1 0.02 1 0.005 0.006 0.027 0.018 0.015 0.023 0.018
p-value * LFO vs. NED 0.0070 NS NS NS NS NS 0.0033 NS NS
* p-values determined by Student’s t-test with p = 0.01 as level of significance; LFO = Local failure only; NED = No evidence of disease.
Group All patients External beam I-125 implant RI B2
C WD MD PD
LFO 0.33 0.35 0.26 0.22 0.40 0.30 0.33 0.32 0.36
+ 0.17 rfr 0.19 + 0.08 f 0.08 + 0.20 t 0.16 + 0.10 + 0.20 t 0.21
cell nucleus
NED 0.56 0.57 0.51 0.63 0.45 0.64 0.60 0.55 0.51
f 0.21 Ik 0.25 + 0.16 k 0.24 * 0.22 + 0.23 + 0.25 + 0.27 + 0.17
p-value * LFO vs. NED 0.0039 NS NS NS NS 0.0044 NS NS NS
* p-values determined by Student’s t-test with p = 0.01 as level of significance; LFO = Local failure only; NED = No evidence of disease.
434
1. J. Radiation Oncology 0 Biology 0 Physics
values compared with matched tumors which responded favorably (NED). The elevated tumor NRF values in the LFO group appeared in those patients with well-differentiated tumors (Table 4). This probably reflects a higher degree of nuclear pleomorphism in those WD cases who failed locally without distant disease. Thus, from a prospective view, there may be a cohort in those patients with WD tumors whose elevated tumor NRF values portend an unfavorable local response to radiation therapy. Nuclear roundness factor values were also higher in the MD and PD tumors of those who failed locally compared with MD and PD tumors which were NED. The differences were not significant, and may be more difficult to demonstrate because the NRF of the NED MD and PD
Volume 24, Number 3, 1992
tumors were elevated ( 1.063 and 1.078) compared with NED WD tumors (1.05 1). This may reflect the fact that increasing grade by itself is consistent with elevated NRF values. It would be of interest to correlate tumor NRF with other markers, such as DNA content and indices of cell proliferation, for example, the percent of S-phase cells. Attempts are currently underway to re-analyze slides of these same patients for DNA content, using computerassisted image analysis of microdensitometry patterns following Feulgen staining. In this way, it can be determined whether or not NRF is an independent marker for local failure following radiotherapy, compared with DNA content.
REFERENCES
1. Barlogie, B.; Johnston, D. A.; Smallwood, L.; Raber, M. N.; Maddox, A. M.; Latreille, J.; Swartzendruber, D. E.; Drewinko, B. Prognostic implications of ploidy and proliferative activity in human solid tumors. Cancer Genet. Cytogenet. 6: 17-28; 1982. 2. Benson, M. C.; McDougal, D. C.; Coffey, D. S. The use of multiparameter flow cytometry to assess tumor cell heterogeneity and grade prostatic cancer. Prostate 5:27-45; 1984. 3. Backing, A.; Aufferman, W.; Schwarz, H.; Bammert, J.; Dorrjer, G.; Vucicuja, S. Cytology of prostatic carcinoma. Quantification and validation of diagnostic criteria. Anal. Quant. Cytol. 6:74-88; 1984. 4. Brachman, D. G.; Rubin, S. J.; Hallahan, D.; Beckett, M. A.; Ashman, C.; Yandell, D.; Weichselbaum, R. R. Two prostate carcinoma cell lines demonstrate abnormalities in tumor suppressor genes (Abstr.). Int. J. Radiat. Oncol. Biol. Phys. 19(Suppl.):254; 1990. 5. Diamond, D. A.; Berry, S. J.; Jewett, H. J.; Eggleston, J. C.; Coffey, D. S. A new method to assess metastatic potential of human prostate cancer: relative nuclear roundness. J. Ural. 128:729-734; 1982. 6. El-Mahdi, A. M.; Kuban, D. A.; Schellhammer, P. F. The treatment of choice for localized poorly differentiated adenocarcinoma of the prostate. Am. J. Clin. Oncol. (CCT) 8: 477-480;1985. I. Epstein, J. I.; Berry, S. J.; Eggleston, J. C. Nuclear roundness factor. Cancer 54: 1666- 167 1; 1984. 8. Frankfurt, 0. S.; Chin, J. L.; Englander, L. S.; Greco, W. R.; Pontes, J. E.; Rustum, Y. M. Relationship between DNA ploidy, glandular differentiation, and tumor spread in human prostate cancer. Cancer Res. 45:1418-1423;1985. 9. Gleason, D. F.; Mellinger, G. T.; and the Veterans Administration Cooperative Urologic Research Group. The prediction of prognosis for prostatic adenocarcinoma by combined histologic grading and clinical staging. J. Urol. I 11: 58-64; 1974. 10. Gumerlock, P. H.; Poonamallee, U. R.; Meyers, F. J.; deVere White, R. W. Activated ras alleles in human carcinoma of the prostate are rare. Cancer Res. 5 l(6): 1632- 1637; 199 1. Il. Hedley, D. W.; Friedlander, M. L.; Taylor, 1. W. Application of DNA flow cytometry to paraffin-embedded archival material for the study of aneuploidy and its clinical significance. Cytometry 6:327-333;1985. 12. Helman, L. J.; Thiele, C. J. New insights into the causes of cancer. Pediatr. Clin. North Am. 38(2):20 1-22 1; 199 1. 13. Helpap, B. Observations on the number, size and localization of nucleoli in hyperplastic and neoplastic prostatic disease. Histopathology 13:203-2 11; 1988. 14. Kelemen, P. R.; Buschmann, R. J.; Weisz-Carrington, P.
IS.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
Nucleolar prominence as a diagnostic variable in prostatic carcinoma. Cancer 65:1017-1020;1990. Kuban, D. A.; El-Mahdi, A. M.; Schellhammer, P. F. I-125 interstitial implantation for prostate cancer. What have we learned 10 years later? Cancer 63:24 15-2420; 1989. Lundberg, S.; Carstensen, J.; Rundquist, I. DNA flow cytometry and histopathological grading of paraffin-embedded prostate biopsy specimens in a survival study. Cancer Res. 47:1973-1977;1987. Mettlin, C.; Lee, F.; Drago, J.; Murphy, G. P.; and the Investigators of the American Cancer Society National Prostate Cancer Detection Project. The American Cancer Society National Prostate Cancer Detection Project. Findings on the detection of early prostate cancer in 2,425 men. Cancer 67:2949-2958; 199 1. Miller, G. J.; Shikes, J. L. Nuclear roundness as a predictor of response to hormonal therapy of patients with stage D2 prostatic cancer. In: Karr, J. P., Coffey, D. S., Gardner, W., eds. Prognostic cytometry and cytopathology of prostate cancer. New York, NY: Elsevier; 1988:349-353. Mohler, J. L.; Partin, A. W.; Coffey, D. S. Cancer cell motility: a visual grading system for assessment of prognosis in prostatic carcinoma. In: Karr, J. P., Coffey, D. S., Gardner, W., eds. Prognostic cytometry and cytopathology of prostate cancer. New York, NY: Elsevier; 1988;337-348. Mohler, J. L.; Partin, A. W.; Coffey, D. S. Correlation of prognosis to nuclear roundness and flow cytometric light scatter. Anal. Quant. Cytol. Histol. 9: 156-164;1987. Paulson, D. F.; Stone, A. R.; Walther, P. J.; Tucker, J. A.; Cox, E. B. Radical prostatectomy: anatomical predictors of success or failure. J. Urol. 136:1041-1043;1986. Pilepich, M. V.; Krall, J. M.; Sause, W. T.; Johnson, R. J.; Russ, H. H.; Hanks, G. E.; Perez, C. A.; Zinninger, M.; Martz, K. L. Prognostic factors in carcinoma of the prostate-analysis of RTOG study 75-06. Int. J. Radiat. Oncol. Biol. Phys. 13:339-349;1987. Stephenson, R. A.; James, B. C.; Gay, H.; Fair, W. R.; Whitmore, W. F., Jr.; Melamed, M. R. Flow cytometry of prostate cancer: relationship of DNA content to survival. Cancer Res. 47:2504-2509;1987. Treiger, B.; Isaacs, J. Expression of a transfected v-Harveyras oncogene in a Dunning rat prostate adenocarcinoma and the development of high metastatic ability. J. Urol. 140(6):1580-1586;1988. Viola, M. V.; Fromouritz, F.; Oravez, S.; Deb, S.; Finkel, G.; Lundy, J.; Hand, P.; Thor, A.; Schlom, J. Expression of ras oncogene p 21 in prostate cancer. New England J. Med. 314:133-137;1986.
24, pp.
0360.3016/92 $5.00 + .oO Copyright 8 1992 Pergamon Press Ltd.
431434
??Clinical Original Contribution
NUCLEAR ROUNDNESS FACTOR AND LOCAL FAILURE FROM DEFINITIVE RADIATION THERAPY FOR PROSTATIC CARCINOMA JAMES SHAEFFER, CHRISTINE
PH.D.,
B. PHILPUT,
JAMES A. TEGELER, PH.D.
AND ANAS
M.D.,
DEBORAH
M. EL-MAHDI,
A. KUBAN,
M.D.,
Sc.D.,
M.D., FACR
Department of Radiation Oncology and Biophysics, Eastern Virginia Medical School, Norfolk, VA Of 375 patients with prostatic carcinoma treated definitively with radiation therapy at this institution with at least a 5 year follow-up, 23 patients failed locally only, 72 failed with distant metastasis only, 60 had both local and distant failure, while 220 showed no evidence of disease. In search for a possible marker for local failure following radiation therapy, we examined several nuclear morphometric parameters which have been shown to correlate with the biologic aggressiveness of this disease. The 23 locally failed only patients were matched with 23 no evidence of disease patients for stage, grade, treatment modality, prior surgery, age at diagnosis and race. Archival hematoxylin and eosin slides were obtained for 22 of the 23 matched pairs, and morphometric features, including nuclear roundness factor and nuclear area, as well as numbers of nucleoli were assessed using computer-assisted image analysis in both tumor cells and normal prostatic epithelium. Tumor nuclei from the locally failed only patients had significantly higher nuclear roundness factor values (p = 0.0089) compared with tumor cells from no evidence
of disease patients. Analysis of these data by clinical stage demonstrated no significant differences between the locally failed only and no evidence of disease patients. Likewise, there were no significant differences in nuclear roundness factor values of locally failed only and no evidence of disease patients with poorly or moderately welldifferentiated tumors. However, there was a highly significant difference (p = 0.0012) in the nuclear roundness factor values of locally failed only and no evidence of disease patients with well-differentiated tumors. Thus, there appears to be a subset of patients with well-differentiated adenocarcinoma of the prostate who have significantly more irregular tumor nuclei and who fail locally only following definitive radiation therapy. Prostate carcinoma, Radiotherapy,
Nuclear roundness factor, Local failure.
categorize the variable biological behavior of prostate cancer. Nuclear morphology (5, 7, 18, 20, 2 I), motility ( 19) DNA content and various indices of cellular proliferation ( 1,2,8, 11, l&23), growth factor (I 2) oncogenes
INTRODUCTION Prostatic cancer occurs frequently and the incidence will most likely continue to rise in the 1990’s and beyond because of the “aging” of the U.S. male population
(10, 24, 25) and suppressor genes (4) have been investigated for their ability to serve as markers of patient outcome for prostatic carcinoma. We report here on the correlation of nuclear morphology with clinical outcome in a retrospective study involving patients with prostatic carcinoma given definitive radiation therapy.
and ever-increasing emphasis on screening and early detection ( 17). There is frequently great variation in the biologic behavior of prostatic tumor cells, even when they appear histologically similar. Consequently, the clinical course can be unpredictable. Some patients will have widespread disease within a relatively short time after diagnosis whereas others will have a relatively indolent disease which does not significantly alter life span, and which requires little or no clinical intervention. A major challenge will be to search for and identify prognostic indicators or markers which more accurately
METHODS
AND
MATERIALS
In a series of 375 patients with prostatic carcinoma treated definitively at this institution with radiation therapy and having a minimum of 5 years’ follow-up, the
Presented at the 32nd Annual ASTRO Meeting, Miami Beach, FL, 15- 19 October 1990. Reprint requests to: James Shaeffer, Ph.D., 600 Gresham Dr., Norfolk, VA 23507. Acknowledgements-The authors wish to acknowledge Karen H. Manias for her excellent technical assistance, Therese Babb and Claudia G. Brigham for their efforts in obtaining clinical
information, and Stephanie M. White and Bernard Weintraub for their help in preparing the manuscript. This investigation was supported in part by a grant from the Sentara Cancer Center. Dr. El-Mahdi is a recipient of The American Cancer Society Professorship of Clinical Oncology. Accepted for publication 17 March 1992. 431
432
1. J. Radiation Oncology 0 Biology 0 Physics
clinical results were as follows: 220 (59%) had no evidence of disease (NED); 72 (19%) had distant metastasis only (MET); 60 (16%) had distant metastasis and local failure (MET + LF); 23 (6%) had local failure only (LFO). When analyzed by histologic grade and clinical stage, the results were as shown in Table 1. As in previous studies (6, 15) the rate of local recurrence increased with more advanced histologic grade and clinical stage of the tumor. The total rate of local recurrence may be determined by combining the local failure with metastasis groups (MET + LF) with the local failure only groups (LFO). Poorly differentiated (PD) tumors had significantly greater (p < 0.00 1 by chi-square) rates of total local failure compared with well-differentiated (WD) tumors. Patients with stage AZ disease had significantly lower rates (p < 0.001 by chi-square) of total local failure compared with patients with Stage C disease. Likewise, the rate of distant failure also increased with increasing grade and stage. The total rate of distant failure is the combination of metastasis only groups (MET) with the local failure with distant metastasis groups (MET + LF). Poorly differentiated tumors again had significantly higher rates (p < 0.001 by Chi-square) compared to WD tumors, and both A2 and B, patients had lower rates of distant failure (p < 0.00 1 by Chi-square) compared with Stage C patients. Nuclear morphometric measures have been correlated with the biologic behavior of prostatic carcinoma (5, 7, 18, 20, 21). Likewise, nuclear area (3) and nucleolar prominence (3, 13, 14) have been widely used as diagnostic criteria for prostatic carcinoma. In search of a possible marker for local failure from radiation therapy, we compared nuclear morphometric measures and nucleoli in LFO patients versus a matched set of NED patients. The selection of these two groups eliminated possible confounding factors from distant metastasis. The 23 LFO patients were matched with 23 NED patients on the basis of previously reported prognostic factors (9, 22): ??
Histologic grade: well differentiated (WD), moderately differentiated (MD), or poorly differentiated (PD); Table 1. Metastasis and local failure by grade and stage MET only
MET + LF
LFO
Volume 24, Number 3, 1992
Fig. 1. Illustrative examples of shape and nuclear roundness factor. A perfect circle has a nuclear roundness factor (NRF) value of 1.00. Greater degrees of eccentricity with a smooth surface will increase NRF as shown by nuclei with NRF values of 1.06 and 1.37. Also, roughness of the edge will also contribute to increased NRF values, such as the nuclei with values of I. 18 and 1.24. (Modified from Diamond et al. (5) with permission.)
Clinical stage: B1, B2 or C; ?? Prior surgery, for example, transurethral the prostate (TURP); ?? Age at diagnosis; ?? Race: black, white, other. ??
Patients were also matched by treatment modality, either external beam or interstitial I- 125. Archival hematoxylin and eosin slides were obtained for 22 of the 23 matched pairs. Appropriate areas containing tumor and normal prostatic epithelium were reviewed by the same pathologist (J.A.T.), given Gleason (9) scores as the sum of the primary plus secondary scores, and then “blind” coded. Slides were viewed under a 100X oil immersion lens. Nuclear roundness factor (NRF), nuclear area, and numbers of nucleoli per cell were assessed Table 2. Characteristics
Characteristic Treatment modality External beam I- 125 implant Clinical stage BI J-52
Histologic grade WD MD PD Clinical stage A2
Bl B2 C
No. (%)
No. (W)
No. (%)
1 l/l38 (7.9) 31/134 (23.1) 30/103 (29.1)
IO/l38 (7.2) 22/134 (16.4) 28/103 (27.2)
7/138 (5.1) 10/134 (7.5) 6/103 (5.8)
5/49 3/27 24/146 40/153
2/49 0127 23/146 35/153
O/49 3/27 9/146 1 l/153
(10.2) (11.1) (16.4) (26.1)
(4.1) (0) (15.8) (22.9)
(0) (11.1) (6.2) (7.2)
MET = Distant metastasis; MET + LF = Distant metastasis and local failure; LFO = Local failure only.
resection of
C Histologic grade WD MD PD Prior TURP Yes No Age at diagnosis Race Black White
of the study population LF no. patients
NED no. patients
18 4
18 4
3 8 11
3 8 11
7 10 5
7 11 4
8 14 66.2
8 14 68.4
5 17
2 20
LF = Local failure; NED = No evidence
of disease.
433
NRF and prostatic carcinoma 0 J. SHAEFFERet al. Table 3. Summary
Nuclear
vs. normal area, p* area, CL* vs. normal cell cell vs. normal
morphometric
1.082 + 0.027 1.040 + 0.018 0.000 1 34.18 k 11.69 26.28 + 4.82 0.0078 0.71 f 0.21 0.33 + 0.18 0.000 1
1.062 f 0.021 1.037 + 0.009 0.000 1 35.66 + 10.69 29.90 f 7.14 NS 0.82 + 0.13 0.56 t 0.23 0.0008
of 50 tumor nuclei and 50 normal nuclei using the Zeiss Videoplan image analysis system. Areas of tumor and normal prostatic epithelium were selected by the pathologist (J.A.T.) with care to avoid areas ofedge artifact, crush artifact and cautery artifact. Nuclear roundness factor is defined as (nuclear perimeter/27r) + (nuclear area/?r)0.5. Nuclear roundness factor values for a perfect circle are 1.Oand increase above 1.O as the shape of the nucleus becomes more ellipsoid or if the nuclear membrane is more irregular, as illustrated in Figure 1. on a minimum
RESULTS The clinical characteristics of the local failure and NED patients are presented in Table 2. A summary of the results of nuclear parameters is given in Table 3. In each group (LFO and NED), tumor nuclei were shaped more irregularly (had higher values of NRF), were larger, and had more nucleoli compared with normal prostatic epithelial nuclei. Comparing LFO to NED populations, however, tumor NRF values were elevated in the LFO group, while the NED group had more nucleoli in the normal nuclei. When tumor NRF was further studied by treatment modality, grade, and stage (Table 4), the difference in NRF values between the two groups was significant only in those with well-differentiated tumors. Similarly, a significant nuclear roundness
p-value * LFO vs. NED
NED
* p-values determined by Student’s t-test with p = 0.0 1 as level of significance: disease; NRF = nuclear roundness factor.
Table 4. Tumor
data
LFO
parameter
Tumor NRF Normal NRF p-value, tumor Tumor nuclear Normal nuclear p-value, tumor Nucleoli/tumor Nucleoli/normal p-value, tumor
of nuclear
0.0070 NS NS NS NS 0.0039
LFO = Local failure only; NED = No evidence
difference in nucleoli/normal cell nucleus was demonstrated only in those with Stage C disease (Table 5). DISCUSSION Relative NRF is the absolute NRF value of the tumor cells divided by the absolute NRF of the normal prostatic epithelial nuclei. Relative NRF values may be preferable to absolute NRF values because relative NRF values may minimize various fixation artifacts (20, 5). We used absolute NRF values in our study, however, because suitable fields of normal prostatic epithelium were found in only 3 1 of the 44 cases analyzed. As might be expected (Table 3) tumor cells from both local failure and NED groups had greater numbers of nucleoli/cell than normal cells. This may reflect the greater degree of aneuploidy presumed in the malignant cells compared with normal (diploid) cells. We have no explanation for the observation of significantly greater numbers of nucleoli/cell in the normal cells of NED patients compared with LFO patients. Since differences between nucleoli/cell in NED and LFO patients were seen only in normal, but not tumor cells, a more detailed analysis of the differences based on grade and stage were presented in Table 5. A corresponding table for tumor cells was not presented since the differences were not significant. Tumors which failed locally had higher absolute NRF
factor data
Table 5. Nucleolus/normal Group
LFO
All patients External beam I- 125 implant RI
1.082 + 0.027 1.08 1 t 0.027 1.086 r 0.028 1.080 f 0.025 1.08 1 ‘- 0.024 1.083 t 0.031 1.085 k 0.015 1.074 f 0.03 1 1.095 +- 0.029
B2
C WD MD PD
NED 1.062 1.066 1.043 1.047 1.062 1.066 1.051 1.063 1.078
of
+ + + f f f + f f
0.02 1 0.02 1 0.005 0.006 0.027 0.018 0.015 0.023 0.018
p-value * LFO vs. NED 0.0070 NS NS NS NS NS 0.0033 NS NS
* p-values determined by Student’s t-test with p = 0.01 as level of significance; LFO = Local failure only; NED = No evidence of disease.
Group All patients External beam I-125 implant RI B2
C WD MD PD
LFO 0.33 0.35 0.26 0.22 0.40 0.30 0.33 0.32 0.36
+ 0.17 rfr 0.19 + 0.08 f 0.08 + 0.20 t 0.16 + 0.10 + 0.20 t 0.21
cell nucleus
NED 0.56 0.57 0.51 0.63 0.45 0.64 0.60 0.55 0.51
f 0.21 Ik 0.25 + 0.16 k 0.24 * 0.22 + 0.23 + 0.25 + 0.27 + 0.17
p-value * LFO vs. NED 0.0039 NS NS NS NS 0.0044 NS NS NS
* p-values determined by Student’s t-test with p = 0.01 as level of significance; LFO = Local failure only; NED = No evidence of disease.
434
1. J. Radiation Oncology 0 Biology 0 Physics
values compared with matched tumors which responded favorably (NED). The elevated tumor NRF values in the LFO group appeared in those patients with well-differentiated tumors (Table 4). This probably reflects a higher degree of nuclear pleomorphism in those WD cases who failed locally without distant disease. Thus, from a prospective view, there may be a cohort in those patients with WD tumors whose elevated tumor NRF values portend an unfavorable local response to radiation therapy. Nuclear roundness factor values were also higher in the MD and PD tumors of those who failed locally compared with MD and PD tumors which were NED. The differences were not significant, and may be more difficult to demonstrate because the NRF of the NED MD and PD
Volume 24, Number 3, 1992
tumors were elevated ( 1.063 and 1.078) compared with NED WD tumors (1.05 1). This may reflect the fact that increasing grade by itself is consistent with elevated NRF values. It would be of interest to correlate tumor NRF with other markers, such as DNA content and indices of cell proliferation, for example, the percent of S-phase cells. Attempts are currently underway to re-analyze slides of these same patients for DNA content, using computerassisted image analysis of microdensitometry patterns following Feulgen staining. In this way, it can be determined whether or not NRF is an independent marker for local failure following radiotherapy, compared with DNA content.
REFERENCES
1. Barlogie, B.; Johnston, D. A.; Smallwood, L.; Raber, M. N.; Maddox, A. M.; Latreille, J.; Swartzendruber, D. E.; Drewinko, B. Prognostic implications of ploidy and proliferative activity in human solid tumors. Cancer Genet. Cytogenet. 6: 17-28; 1982. 2. Benson, M. C.; McDougal, D. C.; Coffey, D. S. The use of multiparameter flow cytometry to assess tumor cell heterogeneity and grade prostatic cancer. Prostate 5:27-45; 1984. 3. Backing, A.; Aufferman, W.; Schwarz, H.; Bammert, J.; Dorrjer, G.; Vucicuja, S. Cytology of prostatic carcinoma. Quantification and validation of diagnostic criteria. Anal. Quant. Cytol. 6:74-88; 1984. 4. Brachman, D. G.; Rubin, S. J.; Hallahan, D.; Beckett, M. A.; Ashman, C.; Yandell, D.; Weichselbaum, R. R. Two prostate carcinoma cell lines demonstrate abnormalities in tumor suppressor genes (Abstr.). Int. J. Radiat. Oncol. Biol. Phys. 19(Suppl.):254; 1990. 5. Diamond, D. A.; Berry, S. J.; Jewett, H. J.; Eggleston, J. C.; Coffey, D. S. A new method to assess metastatic potential of human prostate cancer: relative nuclear roundness. J. Ural. 128:729-734; 1982. 6. El-Mahdi, A. M.; Kuban, D. A.; Schellhammer, P. F. The treatment of choice for localized poorly differentiated adenocarcinoma of the prostate. Am. J. Clin. Oncol. (CCT) 8: 477-480;1985. I. Epstein, J. I.; Berry, S. J.; Eggleston, J. C. Nuclear roundness factor. Cancer 54: 1666- 167 1; 1984. 8. Frankfurt, 0. S.; Chin, J. L.; Englander, L. S.; Greco, W. R.; Pontes, J. E.; Rustum, Y. M. Relationship between DNA ploidy, glandular differentiation, and tumor spread in human prostate cancer. Cancer Res. 45:1418-1423;1985. 9. Gleason, D. F.; Mellinger, G. T.; and the Veterans Administration Cooperative Urologic Research Group. The prediction of prognosis for prostatic adenocarcinoma by combined histologic grading and clinical staging. J. Urol. I 11: 58-64; 1974. 10. Gumerlock, P. H.; Poonamallee, U. R.; Meyers, F. J.; deVere White, R. W. Activated ras alleles in human carcinoma of the prostate are rare. Cancer Res. 5 l(6): 1632- 1637; 199 1. Il. Hedley, D. W.; Friedlander, M. L.; Taylor, 1. W. Application of DNA flow cytometry to paraffin-embedded archival material for the study of aneuploidy and its clinical significance. Cytometry 6:327-333;1985. 12. Helman, L. J.; Thiele, C. J. New insights into the causes of cancer. Pediatr. Clin. North Am. 38(2):20 1-22 1; 199 1. 13. Helpap, B. Observations on the number, size and localization of nucleoli in hyperplastic and neoplastic prostatic disease. Histopathology 13:203-2 11; 1988. 14. Kelemen, P. R.; Buschmann, R. J.; Weisz-Carrington, P.
IS.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
Nucleolar prominence as a diagnostic variable in prostatic carcinoma. Cancer 65:1017-1020;1990. Kuban, D. A.; El-Mahdi, A. M.; Schellhammer, P. F. I-125 interstitial implantation for prostate cancer. What have we learned 10 years later? Cancer 63:24 15-2420; 1989. Lundberg, S.; Carstensen, J.; Rundquist, I. DNA flow cytometry and histopathological grading of paraffin-embedded prostate biopsy specimens in a survival study. Cancer Res. 47:1973-1977;1987. Mettlin, C.; Lee, F.; Drago, J.; Murphy, G. P.; and the Investigators of the American Cancer Society National Prostate Cancer Detection Project. The American Cancer Society National Prostate Cancer Detection Project. Findings on the detection of early prostate cancer in 2,425 men. Cancer 67:2949-2958; 199 1. Miller, G. J.; Shikes, J. L. Nuclear roundness as a predictor of response to hormonal therapy of patients with stage D2 prostatic cancer. In: Karr, J. P., Coffey, D. S., Gardner, W., eds. Prognostic cytometry and cytopathology of prostate cancer. New York, NY: Elsevier; 1988:349-353. Mohler, J. L.; Partin, A. W.; Coffey, D. S. Cancer cell motility: a visual grading system for assessment of prognosis in prostatic carcinoma. In: Karr, J. P., Coffey, D. S., Gardner, W., eds. Prognostic cytometry and cytopathology of prostate cancer. New York, NY: Elsevier; 1988;337-348. Mohler, J. L.; Partin, A. W.; Coffey, D. S. Correlation of prognosis to nuclear roundness and flow cytometric light scatter. Anal. Quant. Cytol. Histol. 9: 156-164;1987. Paulson, D. F.; Stone, A. R.; Walther, P. J.; Tucker, J. A.; Cox, E. B. Radical prostatectomy: anatomical predictors of success or failure. J. Urol. 136:1041-1043;1986. Pilepich, M. V.; Krall, J. M.; Sause, W. T.; Johnson, R. J.; Russ, H. H.; Hanks, G. E.; Perez, C. A.; Zinninger, M.; Martz, K. L. Prognostic factors in carcinoma of the prostate-analysis of RTOG study 75-06. Int. J. Radiat. Oncol. Biol. Phys. 13:339-349;1987. Stephenson, R. A.; James, B. C.; Gay, H.; Fair, W. R.; Whitmore, W. F., Jr.; Melamed, M. R. Flow cytometry of prostate cancer: relationship of DNA content to survival. Cancer Res. 47:2504-2509;1987. Treiger, B.; Isaacs, J. Expression of a transfected v-Harveyras oncogene in a Dunning rat prostate adenocarcinoma and the development of high metastatic ability. J. Urol. 140(6):1580-1586;1988. Viola, M. V.; Fromouritz, F.; Oravez, S.; Deb, S.; Finkel, G.; Lundy, J.; Hand, P.; Thor, A.; Schlom, J. Expression of ras oncogene p 21 in prostate cancer. New England J. Med. 314:133-137;1986.
