Radiation Therapy Failure in Prostate Cancer Patients: Risk Factors and Methods of Detection

Abstract

Radiation therapy for clinically localized prostatic carcinoma remains one of the mainstays among therapeutic approaches; however, patients continue to fail radiation therapy at too high a rate. This article reviews the risk factors and methods of detection for prostate cancer recurrence. The relative merits of the three major pre-therapy prognostic indicators—TNM staging, Gleason score, and serum prostate-specific antigen (PSA) levels—are discussed. The use of staging and Gleason score, as well as digital rectal examination, transrectal ultrasound, and post-radiation prostate biopsies in detecting failure of radiation therapy is reviewed. Challenges relating to the use of serum PSA levels as an indicator of recurrence are examined. Finally, this article makes recommendations as to procedure for evaluating patients suspected of failing radiation therapy.

Radiation therapy for clinically localized prostatic carcinoma has remained one of the mainstays among our therapeutic approaches. Indeed, the pendulum is recently shifting to increased use of radiation therapy, primarily by brachytherapy, for a variety of reasons. There have been significant improvements in methodology. For example, ultrasound guidance for brachytherapy appears to enable more precise and accurate volume distribution of the radiation. More sophisticated imaging and computer enhancement modalities for external beam therapy are now also widely utilized. However, patients continue to fail radiation therapy at too high a rate.

Risk Factors for Failure of Radiation Therapy

The etiology of failure to eradicate the cancer is obviously multifactorial. These factors include relative radioresistance of prostatic carcinoma, failure to administer a cytotoxic dose to the entire gland, and limitations in the ability to increase dose owing to potential injury to surrounding structures. A significant possibility is that as prostate cancer is being detected earlier and earlier, men are at risk for progression and/or development of secondary carcinomas for a much greater time.

Clinical recurrence, whether detected as it is most commonly by biochemical evidence of failure based on prostate-specific antigen (PSA) assay, or on clinical grounds, has been demonstrated to predict ultimate disease dissemination.^1,2 It has been demonstrated that patients with local persistance of cancer after definitive radiation therapy have shortened disease-specific survival as well as a four-fold increase for metastasis compared to patients with apparent local control. Morbidity owing to tumor progression may include pain, hematuria, bladder outlet obstruction, ureteral obstruction, and resulting renal insufficiency, and may also be the source for metastasis. Therapy for local disease persistence, including transurethral resection of the prostate (TURP) or other forms of urinary diversion, is commonly required. TURP for bladder outlet obstruction in the setting of radiation failure is associated with significant morbidity, including incontinence in approximately 30% of patients.³

Pre-therapy Indicators of Prostatic Carcinoma

Three major pre-therapy prognostic indicators are widely used. These are the clinical stage (tumor, node, metastasis [TNM]), the Gleason score, and the serum PSA level. Unfortunately, problems are associated with all these. The lack of prognostic information associated with clinical stage is perhaps best exemplified by the study by Pisansky et al.⁴ He reported on 500 patients with external beam radiation therapy. Twenty-three percent of patients who at diagnosis were clinical stage T2b had biochemical recurrence within 5 years. It should be noted that within subgroups of this cohort (based on grade and pretreatment PSA level), the risk of failure ranged from 0% to 88%. Thus, the importance of tables such as those developed by Partin and associates and others,^5–8 which combined these three important prognostic factors.

Subjectivity of palpation affording clinical stage determination, and indeed of the Gleason score, makes these two variables less robust when subjected to statistical testing analysis. In a number of series,^9–11 when subjected to multivariate analysis, Gleason score and clinical stage were no longer significant. The objectivity associated with serum PSA remains predictive when subjected to multivariate analysis.

Despite tremendous effort in establishment of new markers based on molecular, cytologic, or tissue findings, the heterogeneity observed in the prostate makes the accuracy of these problematic. This is depicted in Figure 1. New markers, such as expression of tumor suppression genes, the reverse transcriptase chain reaction, DNA ploidy, and neovascularity resulting from angiogenesis, are all undergoing extensive scrutiny. Unfortunately, none of these has been demonstrated to provide unique independent prognostic information in widespread clinical series.

Figure 1.

Types of heterogeneity of expression in prostate cancer. Reproduced from Belldegrun A, Kirby RS, Newling DWW, editors. New Perspectives in Prostate Cancer, 2nd edition. Oxford, UK: Isis Medical Media, 2000, with permission from the publisher.

Because of the heterogeneity and difficulty in establishing reliable markers based on biopsy material, most physicians have resorted to utilization of systemically expressed markers, the most widely utilized of course being PSA. Table 1 illustrates the prognostic information afforded by pretreatment, serum PSA levels in men treated with external beam radiation therapy. As is demonstrated, staggeringly different prediction of biochemical disease-free rate is observed across these six large series based on PSA level.^8,9,12–15 Vicissitudes of determining the definition of biochemical failure following radiation therapy will be discussed below. It should be noted that the definition utilized in these papers varies tremendously. Kuban defined it as a PSA greater than 4.0 ng/mL,¹² whereas Zeitman’s definition is PSA greater than 1.0 ng/mL 2 years or more following radiation therapy or an increase of greater than 10% in the first 2 years.¹³

Table 1.

Biochemical Disease-Free Rates at 3 to 5 Years in Men Treated with Radiation therapy as a Function of Pre-therapy PSA Level

	Author/Reference
PreTx PSA	Kuban et al12	Crook et al18*	Zietman et al13	Hanks et al14	Lee et al9	Preston et al15
(ng/mL)	5-y bNED	5-y bNED	4-y bNED	5-y bNED	3-y bNED	5-y bNED
	(N = 652)	(N = 498)	(N = 161)	(N = 502)	(N = 500)	(N = 371)
0–4	69%	90%	81%	83%	85%	79%
4.1–10	58%	62%	43%	-	81%	67%
10.1–20	57%	26%	31%	27%	59%	57%
20.1–50	20%	18%	6%	13%	35%	27%
>50	-	-	0	-	-	0

^*Pretreatment PSA groupings 0–5 ng/mL, 5.1–10 ng/mL, 10,1–20 ng/mL, and >20 ng/mL.

PreTx PSA, pre-treatment prostate-specific antigen level; bNED, no evidence of disease on the basis of the PSA value.

Detection of Radiation Failure

The suspicion of persistent carcinoma following definitive radiation therapy can be raised in many ways. Clinically, it may present with signs and symptoms of bladder outlet obstruction or upper tract obstruction. In historical series it was often detected by an abnormality on digital rectal examination (DRE). The fibrosis and other changes that occur commonly following radiation therapy make DRE relatively crude in assessing persistent disease. Indeed, it has been demonstrated that between 20% and 80% of patients with positive post-radiation biopsies may indeed have a normal DRE.^16,17 In general, DRE is only definitive in the setting of extensive neoplasm. In recent times, the most common method for detecting the potential of radiation failure is rising PSA levels or failure to achieve a significant PSA nadir.

Zagars¹¹ combined clinical stage, serum PSA level, and Gleason score to afford 6 risk categories. Relapse rate among 938 men treated with external beam radiation therapy ranges from 6% for the lowest from this group to 88%, as shown in Table 2.

Table 2.

Prognostic Catagories of Zagars et al¹¹ Based on Multivariate Regression Analysis of 938 Men

Category	T stage	PSA (ng/mL)	Gleason Score	Relapse Rate (%)
1	T1/T2	<4.0	2–6	6
2	T1/T2	≤4.0	7–10	30
		4.1–10.0	2–7
3	T1/T2	4.1–10.0	8–10	40
		10.1–20.0	2–7
4	T3/T4	<10		46
5	T3/T4	10.1–20	2–7	57
6	any T	>20.0	Any Gleason	88
		10.1–20.0	8–10

The grading system devised by Gleason and colleagues has been shown in all studies to be the best predictor of pathologic stage.⁵ However, the useful information represented in the Gleason system occurs primarily at the poles—those with Gleason scores less than 4 or greater than 6. Stamey and associates were the first to show that the percentage of Gleason grade 4 and 5 carcinoma is the best predictor of all in the Stanford series of radical prostatectomies.¹⁸ They noted that 89% of men were without biochemical failure following radical prostatectomy when less than 10% of the specimens revealed Gleason grade 4 or 5 carcinoma. In contrast, 88% of men progressed when 41% or more of their malignancy was high grade. Extending this to patients treated with radiation therapy can be afforded by substantiating studies by Conrad and associates¹⁹ and Wills and colleagues,²⁰ which demonstrated that the number of biopsies exhibiting Gleason 4 and 5 pattern of carcinoma was the best independent predictor of pelvic lymph node extension.

Transrectal Ultrasound Following Radiation Therapy

The use of transrectal ultrasound as a method of identifying persistent carcinoma has been debated. Initial studies demonstrated that the classic hypoechoic peripheral zone pattern persists after radiation, and areas with successful eradication of tumor become isoechoic.^21,22 This has not been the universal experience. Kabalin and colleagues demonstrated lack of specificity of the sonographic appearance in post-radiation biopsies.¹⁶ The use of transrectal ultrasound to guide biopsies has been shown by Egawa²³ to be important. He noted a detection rate in digitally guided biopsies of only 22% as compared to 67% detection of persistent carcinoma in those men undergoing sonographic biopsy guidance.

The Role of Post-Radiation Prostate Biopsies

Prostatic biopsies following radiation therapy have been a standard procedure for a number of years. The ratio of positive biopsy varies tremendously between series (Table 3). Advances in biopsy technique, including transrectal ultrasound guidance and the spring-loaded biopsy devices, have considerably lessened biopsy-associated morbidity. The rationale for post-radiation biopsies is that microscopic evidence of disease persistence will occur at an earlier stage than clinical manifestation of same; earlier detection enables earlier institution of salvage therapy.^24–26 Owing to the relatively cell-cycle-specific nature of radiation injury to carcinoma, the timing of radiation biopsy remains ill-defined. Indeed, it had previously been proposed that biopsies performed within 18 months of radiotherapy were of little biological significance, given that subsequent tumor kill could occur.²⁷ However, in vitro studies have demonstrated the viability of post-radiated prostate cancer cells. Despite these factors, most reports have suggested that men with a positive biopsy post-radiation therapy have a significantly worse prognosis than those with negative biopsies (Table 4).

Table 3.

Results of Post-Radiation Therapy Prostatic Biopsies

Author/Reference	Technique	N	Positive Biopsy (%)
Sewell et al35	EBRT	16	62
Cosgrove et al26	EBRT	9	56
Kurth et al36	EBRT	23	61
Nachstein et al37	EBRT	29	52
Lytton et al38	Brachytherapy	22	50
Kiesling et al39	EBRT	68	57
Jacobi et al40	EBRT	64	39
Scardino et al52	Brachytherapy +	115	33
	EBRT
Babaian et al41	EBRT	31	71
Freiha et al42	EBRT	64	61
Bosch et al43	EBRT +	29	52
	Brachytherapy
Scardino et al17	Brachytherapy +	124	35
	EBRT
Schellhammer et al44	EBRT or	126	33
	Brachytherapy
Bagshaw et al45	EBRT	64	61
Kabalin et al16	EBRT	27	93
Dugan et al46	EBRT	37	38
Kuban et al47	EBRT or	94	18
	Brachytherapy
Goad et al48	EBRT +	49	55
	Brachytherapy
Crook et al49	EBRT	226	30
Kaye et al50	Brachytherapy	71	18
			(24% indeterminate)
Radge et al51	Brachytherapy	77	5
			(13% indeterminate)

EBRT, external beam radiation therapy.

Table 4.

Prognostic Significance of Post-Radiation Therapy Prostatic Biopsy

Author/Reference	Technique	N	Positive Biopsy	Negative Biopsy
Cosgrove et al26	EBRT	9	60% recurrence free at 5 years	100% recurrence free at 5 years
Kurth et al36	EBRT	23	86% recurrence free at 1–4 years	100% recurrence free at 1–4 years
Lytton et al38	Brachytherapy	22	91% recurrence free at 2 years	91% recurrence free at 2 years
Kiesling et al39	EBRT	68	72% progression free in 5 years	96% progression free in 5 years
Jacobi et al40	EBRT	64	36% recurrence free at 4 years	82% recurrence free at 4 years
Scardino52	Brachytherapy +	115	50% recurrence free in	86% recurrence free in 10–24 months
	EBRT		10–24 months
Babaian et al41	EBRT	31	89% with abnormal DRE	11% with abnormal DRE
			64% with normal DRE	36% with normal DRE
Freiha et al42	EBRT	64	28% progression free in 4.5 years	76% progression free in 4.5 years
Scardino et al17	Brachytherapy +	124	35% recurrence free at 5 years	69% recurrence free at 5 years
	EBRT
Schellhammer et al44	EBRT or	126	Actuarial disease free survival	Actuarial disease free
	Brachytherapy		35% at 5 years	survival 85% at 5 years
			Actuarial local failure rate	Actuarial local failure
			50% at 5 years	rate 8% at 5 years
Bagshaw et al45	EBRT	64	28% disease free survival	76% disease free survival at 15 years
			at 15 years
Kabalin16	EBRT	27	100% with abnormal DRE	-
			91% with normal DRE
Kuban et al47	EBRT or	94	Actuarial disease free survival	Actuarial disease free survival
	Brachytherapy		32% at 5 years, 18% at 10 years	82% at 5 years, 62% at 10 years
			Actuarial local failure rate	Actuarial local failure rate
			44% at 5 years, 75% at 10 years	8% at 5 years, 24% at 10 years
Goad et al48	EBRT +	49	70% progression free in	95% progression free in 29.5 months
	Brachytherapy		29.5 months
Crook et al49	EBRT	226	-	93% local progression free at
				36 months

EBRT, external beam radiation therapy; DRE, digital rectal examination.

Despite these observations, postradiation biopsy does not always predict eventual clinical recurrence. Approximately 20% of patients who have post-radiation biopsies will have no clinical evidence of disease at follow-up extending up to 10 years.^17,28 Whether these biopsies either detected latent carcinoma or indeed were misinterpreted, “radiation atypia” remains obscure.

Also of considerable importance is the fact that up to 30% of patients who have local or distant recurrence do so in the face of negative postradiation prostate biopsies. Some of these may have had occult micrometastasis at the time of presentation, and sampling errors in the biopsy procedure are notorious. The interpretation of post-radiation biopsies is one of the most difficult tasks of the surgical pathologist. Radiation atypia in benign prostate glands may mimic carcinoma.²⁹ It has been demonstrated that utilizing immunohistochemical techniques with basal cell-specific keratin monoclonal antibodies is useful in differentiating benign glands from malignant ones. This is because only the benign glands display basal cell immunoreactivity.^30,31

PSA in the Detection of Radiation Failure

The definition of biochemical failure following definitive radiation therapy remains obscure. In men treated with radical prostatectomy, theoretically serum PSA should be zero or at the low-end sensitivity of the assay. The greatest variable is assay and laboratorian reliability in establishing the low-end sensitivity range. Although in theory extra-prostatic sites that elaborate a protein similar to PSA may cause false-positive test results, this does not appear to be relevant in clinical practice.

Unfortunately, in the radiationtreated patients the scenario is considerably different. Not only is there an ongoing effect during the first several years of radiation injury to both the cancer and the benign prostatic elements, but in most series remaining benign elements that appear immunohistochemically to elaborate PSA are present. Evidence that new malignant cells may be associated with persistent PSA following radiation therapy may be found in the report by Willet and colleagues,³² who studied 36 men who had prostate radiation due to treatment for nonprostatic pelvic malignancies. Dose ranged from 45 to 65 Gy, and the mean PSA 3 years following radiotherapy was 0.6 ng/mL. However, 20% of the patients had readings greater than 1.5 ng/mL. No man had a clinical diagnosis of prostate carcinoma. Because of these issues, a number of definitions for biochemical failure following radiation have been utilized. This may well be the biggest factor in explaining the staggeringly different rates of biochemical failure in reported series (Table 5). In 1997, the American Society of Therapeutic Radiation and Oncology (ASTRO) developed a consensus guideline.³³

Table 5.

Actuarial Freedom from PSA and Clinical Failures Based on Definition of a PSA Nadir

Author/Reference	Technique	N	Definition of Failure	Actuarial Freedom	Actuarial Freedom
				from PSA failure	from Clinical Failure
Goad et al48	Brachytherapy +	76	2 consecutively rising PSA	66% at 29.5 months	40% at 5 years
	EBRT		levels or a single rise > 2 ng/mL
Kaplan et al53	EBRT	117	Increasing PSA level	36% at 5 years	49% at 5 years
			after treatment
Schellhammer et al54	EBRT	311	PSA >0.5 ng/mL	30% at 5 years	55% at 5 years
				6% at 10 years	33% at 10 years
	Brachytherapy	123		67% at 5 years	73% at 5 years
				24% at 10 years	52% at 10 years
Stamey et al55	EBRT (98) and	113	PSA > 1 ng/mL	20% at 5 years	-
	Brachytherapy (15)
Zagars et al56	EBRT	314	2 or more consecutive rising	38% at 4 years	-
			PSA level or a second value
			higher than its predecessor by
			1 ng/mL or factor of 1.5
Kavadi et al57	EBRT	182	PSA >1 ng/mL	88% at 5 years	89% at 5 years
Zietman et al13	EBRT	161	PSA < 1 ng/mL or increase by	26% at 48 months	67% at 48 months
			> 10% within the first 2 years
Blasko et al58	Brachytherapy	197	PSA > 4 ng/mL or 2 consecutive	93% at 5 years	-
			PSA level rises or PSA
			> pre-implant value
Critz et al59	Brachytherapy +	239	2 consecutive PSA levels	74% at 5 years	92% at 5 years,
	EBRT		progressively greater than	66% at 10 years	84% at 10 years
			the lowest reading
Kaye et al50	Brachytherapy	45	PSA > 4 ng/mL	98% at 26.3 months	51% at 26.3 months
	EBRT +	31		95% at 26.3 months	63% at 26.3 months
	Brachytherapy
Kuban et al12	EBRT	652	PSA = 4 ng/mL	35% at 5 years	43% at 5 years
				13% at 10 years	31% at 10 years
Rosenzweig et al60	EBRT	285	PSA > 4 ng/mL	53% at 5 years	60% at 5 years
				28% at 10 years	42% at 10 years
Zagars et al61	EBRT	461	2 or more consecutive rising	70% at 5 years	-
			PSA levels or a second value
			higher than its predecessor by
			1 ng/mL or factor of 1.5
Zagars et al11	EBRT	707	2 or more consecutive rising	66% at 5 years	67% at 5 years
			PSA levels or a second value
			higher than its predecessor by
			1 ng/mL or factor of 1.5
Lee et al34	EBRT	364	PSA > 1.5 ng/mL or 2	56% at 5 years	-
			consecutive PSA elevations
Stock et al62	Ultrasound-guided	97	2 consecutive increases in	76% at 2 years	-
	transperineal seed		PSA above a nadir
	implant
Crook et al63	EBRT	207	PSA ≥ 2 ng/mL and 1 ng/mL	82% at 36 months	75% at 36 months
			above the precedent value
Hanks et al64	Conformal 3-D	456	2 consecutive increase in PSA	61% at 5 years	-
	radiation therapy		levels that equals or exceeds	57% at 7 years
			1.5 ng/mL
Radge et al51	Brachytherapy	126	PSA progression or failure	79% at 7 years	77% overall
			to attain PSA value of 1.0 or		7-year survival
			0.5 ng/mL

PSA, prostate-specific antigen; EBRT, external beam radiation therapy.

The ASTRO consensus conference conclusion was that three consecutive rises in PSA is a reasonable definition of biochemical failure. They recommended that determinations be made 3 to 4 months apart in the first 2 years following radiation therapy and every 6 months thereafter. Three sources exist for PSA in patients following radiation therapy—residual benign prostatic epithelium, viable cancer within the prostate, and disseminated disease. Crook has summarized the typical response to therapy according to PSA nadir, time to nadir, and PSA doubling time. This is depicted in Table 6. Time to PSA nadir has been shown to correlate inversely with disease-free survival. Lee and colleagues³⁴ noted that 75% of men whose PSA began to rise following the nadir in less than 12 months had disseminated disease, compared to only 25% of those who achieved their nadir more than 1 year following external beam radiation therapy. PSA doubling times tend to be longer in patients failing locally as compared to those with metastatic disease. Freedom from biochemical relapse at 3 to 5 years according to PSA nadir is shown in Table 7. It is obvious from this that the level of the PSA nadir and the length of nadir provide important prognostic information.

Table 6.

Failure Pattern According to Level of PSA Nadir, Time to Nadir, and PSA Doubling Time

	PSA Nadir	Time to Nadir	PSA Doubling Time
	(ng/mL)	(months)	(months)
NED	0.4–0.5	22–33	NA
Local failure	2.0–3.0	17–20	11–13
Distant failure	5.0–10.0	10–12	3–6

PSA, prostate-specific antigen; NED, no evidence of disease.

Data from Crook et al.⁶⁵

Table 7.

Freedom from Biochemical Relapse (bNED) at 3–5 Years After Radiotherapy According to PSA Nadir

Author / Reference
	Crook et al8	Kestin et al66	Zietman et al67	Lee et al34
PSA nadir	(N = 489)	(N = 871)	(N = 314)	(N = 364)
<0.5	75%	78%	90%	93%
0.6–1.0	50%	60%	55%
1.1–1.9	32%	50%	34%	49%
2.0–3.9	0	20%		16%
>4		9%

bNED, no evidence of disease on the basis of PSA value; PSA, prostate-specific antigen.

Patient Evaluation

Patients suspected of having persistent prostate cancer following radiation therapy owing to either an abnormality on DRE, clinical manifestation, or rising PSA should undergo an ultrasound-guided prostate needle biopsy. Although the best approach remains debatable, we currently perform 10 biopsies including 6 systematic sector biopsies in the parasaggital plane and 2 additional biopsies extended extremely laterally in the gland to sample the so-called anterior horn region. In addition, metastatic workup is mandatory if salvage therapy is considered. This should include bone scan and computed tomography or magnetic resonance imaging of the abdomen and pelvis along with chest radiographs. The use of the ProstaScint scan remains controversial in this setting. The clinician should be cautioned that, whereas it is unusual in the pretreatment setting to have metastatic disease in the absence of PSA greater than 10.0 ng/mL, in the post-radiation cohort this cutoff does not apply. Cystoscopy may be useful to evaluate bladder or bladder neck involvement.

Conclusion

Patients electing radiation therapy need to be followed indefinitely for treatment failure. Risk assessment currently utilizes similar pretreatment tests as with other treatment approaches, including histologic grade, clinical stage, and serum PSA level. Follow-up should include serial DRE and, most importantly, PSA determination. Currently the ASTRO definition for biochemical failure is most widely employed.

Main Points.

• Factors in the failure to eradicate prostate cancer include relative radioresistance of prostatic carcinoma, failure to administer a cytotoxic dose to the entire gland, and limitations in the ability to increase dose owing to potential injury to surrounding structures.

• Three major pre-therapy prognostic indicators are widely used: clinical stage (tumor, node, metastasis [TNM]), the Gleason score, and the serum prostate-specific antigen (PSA) level. Subjectivity of palpation affording clinical stage determination and of Gleason score makes these two variables less robust when subjected to statistical testing analysis; the objectivity associated with serum PSA remains predictive when subjected to multivariate analysis.

• Persistent carcinoma following definitive radiation therapy may present with signs and symptoms of bladder outlet or upper tract obstruction. In historical series it was often detected by an abnormality on digital rectal examination (DRE), but DRE is relatively crude in assessing persistent disease.

• The Gleason grading system has been shown in all studies to be the best predictor of pathologic stage; however, the useful information represented in the Gleason system occurs primarily at the poles—in those with Gleason scores less than 4 or greater than 6.

• Prostatic biopsies following radiation therapy have been a standard procedure for a number of years, and advances in biopsy technique, including transrectal ultrasound guidance and the spring-loaded biopsy devices, have considerably lessened biopsyassociated morbidity.

• Of considerable importance is the fact that up to 30% of patients who have local or distant recurrence do so in the face of negative post-radiation prostate biopsies.

• Regarding serum PSA levels and disease recurrence, the American Society of Therapeutic Radiation Oncology consensus conference conclusion was that three consecutive rises in PSA is a reasonable definition of biochemical failure.

• Patients suspected of having persistent prostate cancer following radiation therapy should undergo an ultrasound-guided prostate needle biopsy. In addition, metastatic workup is mandatory if salvage therapy is considered, including bone scan and computed tomography or magnetic resonance imaging of the abdomen and pelvis along with chest radiographs.