The serum prostate-specific antigen (PSA) level after definitive treatment for prostate cancer (PC) is a powerful predictor of outcome. Occasionally, PC progression can occur despite low or undetectable PSA levels. The authors report on the clinical and pathologic characteristics of patients who experienced PC progression with undetectable or low PSA levels.
From an electronic database of all patients with PC who were treated at The University of Texas M. D. Anderson Cancer Center between 1999 and 2004, a group of 46 patients was identified who had progression to metastatic PC detected with concomitant PSA levels from 0.1 ng/mL to 2 ng/mL. Patient charts were reviewed for tumor stage, Gleason score, pretreatment PSA level, and the presence of atypical histologic variants (ie, ductal, sarcomatoid, or small cell cancers). The nadir PSA level after treatment and the PSA level at the time metastatic PC was detected were determined. The patients were followed semiannually, and imaging studies were obtained at the discretion of treating physicians. The sites of metastasis and histologic confirmation were reported when available.
Twenty-three of 46 patients underwent radical prostatectomy, 11 patients received radiation therapy, and 12 received hormone treatment as their initial form of therapy. Progression to metastatic disease with concomitant, undetectable PSA levels occurred in 10 patients, including 3 patients who had not received treatment with hormones. The sites of metastasis included bone (n = 35 patients), liver (n = 7 patients), retroperitoneal lymph nodes (n = 5 patients), lungs (n = 4 patients), and brain (n = 1 patient). Aggressive and locally advanced PC were common features in these patients: Eighty-five percent had Gleason scores ≥7, 63% had clinical T3 or T4 tumors, and 41% had pretreatment PSA levels >10 ng/mL. Atypical histologic variants were observed in 21 patients (46%) and in 8 of 10 patients who progressed with undetectable PSA levels. In 10 patients (22%), metastasis were detected in the presence of an undetectable PSA level. Eight of those patients had small cell carcinoma. In 19 patients (41%), progression to metastasis occurred without any increase in their PSA from the nadir level. Thirty-one patients (67%) were asymptomatic at the time metastasis was detected, and the detection of metastasis in these patients occurred only because of routine imaging studies.
Progression of PC may occur despite undetectable or low PSA levels. Complete physical evaluation and imaging studies may be indicated in the surveillance of patients with high-grade, locally advanced tumors, especially when atypical histologic variants are present. Cancer 2007. © 2006 American Cancer Society.
Since its introduction in clinical practice in the mid-1980s, prostate-specific antigen (PSA) has become an invaluable tool in the diagnosis and follow-up of patients with prostate cancer (PC). Because PSA expression is limited primarily to the prostate, its presence in the serum usually reflects the existence of prostatic tissue; and, in patients with PC, the serum level of PSA is linked to the burden of cancer.1 The PSA kinetics after treatment of PC is an extremely sensitive marker for disease activity and status. Indeed, biochemical failure after initial therapy for PC is typically the first evidence of disease progression, and it precedes any clinical manifestation of metastatic disease by 5 years to 7 years.2 Similarly, rapid increase in serum PSA levels and a short PSA doubling time have been linked to early disease progression and to a decreased survival rate,3 whereas a persistent undetectable PSA level after radical prostatectomy is associated with prolonged disease-free survival.4 Because of the very high negative predictive value of an undetectable PSA level after radical prostatectomy, such patients usually are considered disease free and do not require any additional staging modalities, such as imaging tests or even digital rectal examination (DRE).5
Progression of PC in the presence of low or undetectable PSA is uncommon and, to date, has been presented in selected case reports.6-8 At the University of Texas M. D. Anderson Cancer Center, we have observed a number of patients with PC who experienced disease progression despite a low or undetectable PSA serum level. The objectives of this report were to describe the phenomenon of PC progression in the presence of a low or undetectable PSA level and to identify potential parameters associated with this particular clinical course.
MATERIALS AND METHODS
The Departments of Genitourinary Medical Oncology and Urology at the University of Texas M. D. Anderson Cancer Center operate an electronic modulary patient database that has included all patients with genitourinary malignancy who presented for treatment to our center since 1999. The database is operated through the Oracle computer operating system, which enables queries based on keywords to identify patient subsets that share specific characteristics of interest.
Abstracting patients' charts into the database was approved by the Institutional Review Board and was in concert with Health Insurance Portability and Accountability Act regulations, and all patients who were included signed informed consent to permit the use of their clinical data for research purposes. In total, 4145 patients who were diagnosed with PC were entered into the database between the years 1999 and 2004. We performed a query of the database using the keywords progression or metastasis in conjunction with serum PSA at the time of progression according to the following PSA cut-off levels: undetectable (<0.1 ng/mL); <0.5 ng/mL, <1 ng/mL, and <2 ng/mL. All serum PSA levels were measured by The University of Texas M. D. Anderson Cancer Center Laboratory using the Bayer Centaur PSA Assay kit (Tarrytown, NY). Disease progression was defined as detection of metastasis in ≥1 of the following sites: bone, viscera, or retroperitoneal lymph nodes above the aortic bifurcation. In this study, involvement of pelvic lymph nodes with PC was not considered as progression. Histologic confirmation of metastatic sites as PC in origin was reported when available. Patients were classified either as hormone naive when they had received no androgen-deprivation therapy (ADT) during at least the 6 months prior to the detection of cancer progression and or hormone treated when they had received ADT during the 6 months prior to disease progression.
After the identification of a subset of patients who had PC progression in the presence of serum PSA levels ≤2 ng/mL, we reviewed each patient chart to obtain the following data: initial clinical T stage, serum PSA level and Gleason score at the time of diagnosis, sites of detected metastatic disease, and imaging modalities applied. The presence of atypical histologic variants, such as neuroendocrine tumors, small cell carcinoma, ductal cancer, and sarcomatoid cancer, was evaluated in all biopsy specimens obtained at the time of diagnosis and also in biopsies from metastatic lesions when these were available.
Patients who were known to harbor another malignancies were excluded from the original group identified by the query to avoid potential bias of misclassifying the origin of disease progression in the presence of a low serum PSA level. Patients in whom the date of detection of metastasis preceded that of the nadir PSA level also were excluded from the analysis.
Patients were classified according to the primary form of treatment received at the time of diagnosis. Twenty-three patients underwent open radical prostatectomy. None of our patients underwent perineal or laparoscopic radical prostatectomy. Eleven patients received radiation therapy as their initial form of therapy. All of those patients received external beam radiation therapy (intensity-modulated radiation therapy). Twelve patients were started on systemic therapy as their initial form of treatment. This included a luteinizing hormone-releasing hormone agonist after a short course of oral antiandrogens in 11 patients and a combination of hormones and chemotherapy in 1 patient.
After treatment, the nadir level of serum PSA was recorded, and all patients were followed with subsequent PSA measurements, physical examinations, and periodic additional imaging studies. The frequency of follow-up visits and the use of imaging studies varied among the patients based on their initial risk of progression, as determined by the tumor grade, disease stage, and initial serum PSA level and by their response to the primary form of treatment (ie, achieving undetectable serum PSA levels in patients who underwent radical prostatectomy). All patients were seen every 6 months or more frequently after definitive treatment. Descriptive statistical methods were used for various variables of interest in this patient population.
Of 4145 patients who were treated for PC at our institution between 1999 and 2004, the initial query identified a total of 100 patients who had disease progression in the presence of a PSA level ≤2 ng/mL. Further investigation of each individual patient chart revealed that 8 patients had a concomitant cancer that may have been the source of metastatic disease in the presence of a low serum PSA level. These patients were excluded. In addition, in 46 other patients, we could not confirm that the detection of metastatic disease chronologically preceded a low PSA level. We were compelled to ascertain that the low PSA level observed at the time that metastatic disease was detected was indeed the special feature we were seeking in these patients and not the result of therapy given because of the newly diagnosed metastasis. Therefore, we also elected to exclude these patients. Finally, we were left with a total of 46 patients who had documented progression of PC in the presence of a PSA level <2 ng/mL and had no other malignancy.
This group of 46 patients represents 1.1% of all patients who presented with PC at our institution during the study period. The clinical characteristics of these patients are listed in Table 1. Overall, 10 patients (22%) had undetectable serum PSA levels, and 30 patients (65%) had PSA levels of ≤1 at the time of progression to metastatic disease (Table 2).
|Variable||No. of patients (%)|
|ADT ± Chemotherapy||XRT||RP||Total|
|Total no.||12 (26)||11 (24)||23 (50)||46|
|Average age [range], y||61 [47–74]||66 [50–76]||60 [50–69]||62 [47–76]|
|Hormone-naive patients||0||0||7||7 (15)|
|PSA at diagnosis|
|0–9.9 ng/mL||5 (42)||8 (73)||10 (43)||23 (50)|
|10–50 ng/mL||4 (33)||2 (18)||8 (35)||14 (30)|
|>50 ng/mL||3 (25)||1 (9)||1 (4)*||5 (11)*|
|T stage at diagnosis|
|T1c||2 (17)||0||0||2 (4)|
|T2||2 (17)||1 (8)||2 (9)||5 (11)|
|T3-T4||5 (42)†||7 (58)‡||17 (74)§||29 (63)|
|Gleason score at diagnosis|
|5–6||0||3 (27)||2 (9)||5 (11)|
|7–8||5 (42)||4 (36)||11 (48)||20 (43)|
|9–10||6 (50)|||4 (36)||9 (39)|||19 (41)|
|Aggressive histologic variant¶||9 (75)||5 (55)||7 (30)||21 (46)|
|Average mo to progression to metastasis [range]#||21 [3–52]||32 [2–154]||35 [3–102]||30 [2–154]|
- ADT indicates androgen-deprivation therapy;±, with or without; XRT, external-beam radiotherapy; RP, radical prostatectomy; PSA, prostate-specific antigen.
- * The PSA level at diagnosis was unavailable in 4 of the patients who underwent radical prostatectomy.
- † The clinical T stage at diagnosis was unavailable in 3 patients.
- ‡ The clinical T stage at diagnosis was unavailable in 4 patients.
- § The clinical T stage at diagnosis was unavailable in 6 patients.
- | The Gleason score at diagnosis was unavailable in 1 patient.
- ¶ Including ductal, small cell, neuroendocrine. and sarcomatoid cancer variants.
- # Calculated from the time of PSA nadir level to the time of metastatic detection.
|Primary treatment||No. of patients (Cumulative %)||Total|
|PSA < 0.1 ng/mL||PSA 0.1–0.5 ng/mL||PSA 0.51–1.0 ng/mL||PSA 1.1–2 ng/mL|
|RP||4 (16)||5 (36)||4 (52)||10 (100)||23|
|Radiation therapy||2 (18)||4 (55)||1 (63)||4 (100)||11|
|ADT||4 (33)||4 (66)||3 (92)||1 (100)||12|
|Total||10 (22)||13 (50)||7 (65)||17 (100)||46|
- PSA indicates prostate-specific antigen; RP, radical prostatectomy; ADT, androgen-deprivation therapy.
Among the 25 patients who had undergone radical prostatectomy as their initial form of therapy, 7 patients were hormone naive at the time of progression to metastatic disease. All other patients had been receiving ADT at the time of progression.
Overall, at the time of progression to metastatic PC, serum PSA levels increased by a median of 0.25 ng/mL (range, 0–1.75 ng/mL) from the nadir value in the entire patient group. In 19 patients (41%), there was no incremental increase in serum PSA levels from the nadir levels at the time of progression.
Increases in serum PSA levels from nadir values at the time metastases were detected are shown in Figures 1-3 for each initial treatment arm. Prior to disease progression, the median PSA doubling time for the entire group was 7.6 months. The average time from diagnosis to the detection of metastatic disease (time to progression) for each treatment modality listed in Table 1.
Atypical histologic variants of PC were identified in 21 of 46 patients, including 9 patients with ductal PC, 8 patients with small cell carcinoma, 2 patients with neuroendocrine tumors, and 2 patients with sarcomatoid tumors. In 12 of these patients, an atypical histologic PC variant was identified at the time of diagnosis; whereas, in the 9 other patients, the presence of atypical histologic variants was noted only in samples that were obtained from metastatic sites at the time of progression, and there was no such previous evidence in the initial prostate biopsy specimen. It is noteworthy that, in all 9 patients whose atypical variants were apparent only at the time of progression, small cell carcinoma was present in 7 patients, and 2 patients had neuroendocrine tumors. Conversely, all patients with ductal PC had this finding in the initial prostate biopsy specimens. In 8 of 10 patients who progressed with undetectable PSA levels, small cell carcinoma was identified.
The most common sites of metastatic involvement included bones (n = 35 patients), liver (n = 7 patients), retroperitoneal lymph nodes (n = 5 patients), lungs (n = 4 patients), and brain (n = 1 patient). Isolated visceral metastasis in the absence of osseous lesions were identified in 8 patients, including 5 patients with liver metastasis, 2 patients with lung metastasis, and 1 patient with brain metastasis.
The initial imaging modality for detecting metastatic disease was bone scintigraphy (BS) in 27 patients, computerized tomography (CT) or magnetic resonance imaging in 19 patients, and chest x-ray in 1 patient. In another patient, both BS and CT of the abdomen detected metastasis on the same day. Fifteen patients underwent biopsies of metastatic lesions, and metastatic PC was confirmed in all of them.
Symptomatic metastases prompted radiologic imaging studies in 15 patients (33%), including 13 patients who had bone pain as the presenting symptom, 1 patient who had hypercalcemia as the presenting symptom, and 1 patient who presented with neurologic symptoms. The remaining 31 patients (67%) had asymptomatic metastases detected on routine surveillance.
In the current study, we showed that PC progression is possible despite minimal serum PSA elevation. Indeed, 22% of our patients developed metastasis in the presence of an undetectable PSA level. This feature is an uncommon phenomenon and occurred in only 1.1% of all PC patients who presented to our center during the study period. In addition, most patients with low to undetectable PSA levels who progressed to metastatic disease are asymptomatic; therefore, careful surveillance and radiologic imaging should be considered in this patient population, particularly if patients have an aggressive variant (ductal or small cell carcinoma) of PC.
In most patients, PC progression to the metastatic stage is associated with marked PSA elevation; and, among patients with newly diagnosed PC, the likelihood of positive BS or CT is very low with PSA levels <20 ng/mL.9-12 Likewise, among patients who experience biochemical progression after radical prostatectomy, the yield of BS becomes significant only when the PSA exceeds 20 ng/mL and is very low with a PSA level <7 ng/mL.13 In a study of 1916 men, Pound et al. observed that none of the patients developed metastatic disease in the absence of a detectable serum PSA level.5 Consequently, those authors concluded that DRE or imaging studies were not indicated in the presence of an undetectable PSA level.
The widely accepted concept is that PC progression and PSA elevation are associated closely; and, because of its high negative predictive value, all patients with undetectable PSA levels after initial treatment should be considered disease free. Despite this concept, the occurrence of PC progression despite low or undetectable PSA levels has been reported sporadically in case reports and short series.6-8, 14-16
Even among patients who experience biochemical progression after initial therapy, the actual detection of metastasis is rare in the presence of low PSA levels. In a series of 167 patients who had disease progression, only 8 patients with undifferentiated PC developed metastasis with a PSA level <10 ng/mL.17 Conversely, in another report on patients who experienced biochemical progression after radical prostatectomy, the average PSA level in patients who had positive bone scans was 61.3 ng/mL (range, 1.3 ng/mL to 123 ng/mL).18 Although this may be valid for the majority of patients with PC, our series clearly demonstrates that there is a subset of patients with PC in whom disease progression is not heralded by PSA elevation.
The characteristic features of patients with PC who experienced disease progression with low PSA levels included Gleason scores >7, atypical histologic variants, particularly small cell and ductal cancers, and locally advanced tumors. Disease progression despite low PSA levels virtually never occurred in the ordinary low-risk patient (ie, clinical T1c with Gleason scores of 5 or 6).
Others have reported this association between metastatic PC with low PSA levels and atypical histologic variants. Sella et al.19 reported on 18 patients with either neuroendocrine tumors or a combination of adenocarcinoma and small cell PC who had low PSA levels at the time of diagnosis in whom progression to metastatic disease occurred in the presence of a median PSA level of 3 ng/mL. In those patients, other serum markers were positive, including the carcinoembryonic antigen, neuron-specific enolase, CA19-9, and CA-125. In another group of patients with anaplastic PC, the subset of patients with small cell carcinoma progressed to metastatic disease with a median PSA level of 1.6 ng/mL.20
PSA is expressed principally by acinar and ductal epithelial cells and is excreted into the lumen, where it acts to cleave semenogelin I and II in the seminal fluid.21 Because of its specificity to prostatic tissue, the main clinical application of PSA has been in the diagnosis and posttreatment surveillance of PC. The phenomenon of PC progression without concomitant PSA elevation may be explained by proliferation of cell lines that either do not have the capacity of producing PSA (ie, small cell or neuroendocrine tumors) or are poorly differentiated PC cells that have lost their ability to express PSA. A possible link between dedifferentiation of PC cells and decreased PSA expression is reflected by the observation that cancer cells have a lower concentration of PSA compared with benign prostate cells.22
The histogenetic origin of the small cell PC variant is uncertain; however, 3 theories have been implied23: The first sustains that this is a primary malignancy of the neuroendocrine cells, a theory that has been supported by the over expression of neuron-specific enolase, chromogranin A, and other factors, such as bombesin and calcitonin, all of which are expressed normally by cells of the amino precursor uptake and decarboxylation cell family. The second theory suggests that these tumors represent a dedifferentiation of high-grade PC into undifferentiated carcinoma that does not manifest the phenotypic features of PC, such as PSA production and affinity to bone. This theory stems from the observation that many of the small cell PCs coexist with typical PC, and a small cell variant often is found late in the course of the disease that is not present in the specimen at the time of diagnosis. A third theory maintains that the origin of small cell PC is the stem cells, which are pluripotent and may evolve into the typical PSA-secreting tumor versus the undifferentiated small cell variant. Tu et al. proposed that malignancy often originates in stem cells. According to that theory, early progenitor stem cells have a pluripotent capacity of migration, invasion, and incorporation into various distant tissues (metastasis); whereas later, more differentiated phases of stem cells have a more homogenous phenotype and a more limited metastatic potential. Thus, tumors arising from primitive forms of prostatic stem cells may metastasize into diverse organs, such as the lungs, liver, brain, and bone; they do not express PSA but, rather, markers that are either embryonal (carcinoembryonic antigen) or neuroendocrine markers (neuron-specific enolase or chromogranin A). Later, more differentiated forms of stem cells will produce only PSA and will have a preference to bone metastasis.24 Pure small cell PC is a rare subtype and may be explained by either the first or third theory. In such patients, the serum PSA level reflects the bulk and activity of the nonmalignant elements in the prostate and is not the result of tumor growth or progression. More frequently, however, tumors contain both small cell variants and typical PC components. Furthermore, in many patients, the atypical and aggressive variants developed with time and were not present at the time of diagnosis.
Given the biologic heterogeneity of PC and its multifocality, it is possible that, with the higher Gleason grade tumors, there is a greater chance for the occurrence of a clonal shift in the original tumor that may express the features of small cell PC, including biologic aggressiveness and lack of PSA production. This subset of undifferentiated PC with or without atypical histologic variants may exhibit a completely different biologic behavior than that of the garden-variety adenocarcinoma of the prostate.25 This manifests not only with disease progression unaccompanied by PSA elevation but also with a more aggressive nature, greater sensitivity to systemic chemotherapy,20 atypical metastatic sites, and lytic bone lesions. Consequently, such patients may require different surveillance workup, including imaging studies directed at anatomic sites that otherwise are involved infrequently with PC. Routine DRE should be an integral part of physical examination performed during follow-up visits. Although this series focused on the detection of distant metastasis with a concomitant low PSA level, a positive DRE may draw attention to local disease activity. Further study is warranted to glean a better understanding of the biology of these rare cancers.
The clinical implications of our study are significant; and, to our knowledge, this is the largest reported series documenting PC progression with undetectable or low serum PSA levels. In addition, we showed that this specific clinical scenario is more likely in patients who have undifferentiated PC at the time of diagnosis and especially when ductal, neuroendocrine, or small cell cancer variants are present. These findings conflict with the common concepts that PSA is the most sensitive marker for PC progression and that no other studies are necessary when the PSA is low or undetectable. Rather, our findings suggest that, in patients with high-grade PC, especially when atypical PC variants are present, periodic imaging studies should be considered even if the PSA level remains low. In addition, it is possible that, in such patients, the measurement of other serum markers, including neuron-specific enolase, carcinoembryonic antigen, or CA 19-9, may provide early evidence of cancer progression.26
Because of the specific referral patterns to a tertiary cancer center, the subset of patients who experience PC progression with an undetectable or low PSA may have been over represented. It is likely that the true proportion of such patients in the general population of patients with PC is <1%.
In conclusion, progression of PC to metastatic disease can occur despite undetectable or low serum PSA levels. Although this is an uncommon clinical scenario, patients with undifferentiated tumors and those in whom atypical cancer variants are encountered are prone to this event, and imaging studies as well as additional serum tumor markers may be considered even when the PSA level remains low.
Molecular and cellular prostate biology: origin of prostate-specific antigen expression and implications for benign prostatic hyperplasia.
, , , et al.
Natural history of progression after PSA elevation following radical prostatectomy.
, , , et al.
Pattern of prostate-specific antigen (PSA) failure dictates the probability of a positive bone scan in patients with an increasing PSA after radical prostatectomy.
J Clin Oncol.
, , .
Prostatic specific antigen related to clinical status 1 to 14 years after radical retropubic prostatectomy.
Br J Urol.
, , , et al.
Digital rectal examination and imaging studies are unnecessary in men with undetectable prostate specific antigen following radical prostatectomy.
, , , et al.
Progression of prostate cancer despite undetectable serum prostate specific antigen.
Can J Urol.
, , , et al.
Metastatic prostate cancer with normal level of serum prostate-specific antigen.
Int Urol Nephrol.
Prostate cancer with a normal PSA: small cell carcinoma of the prostate—a rare entity.
J Am Board Fam Pract.
, , , et al.
Can prostate-specific antigen levels predict bone scan evidence of metastases in newly diagnosed prostate cancer?
Am J Clin Oncol.
, , , et al.
The clinical value of prostate-specific antigen and bone scintigraphy in the staging of patients with newly diagnosed, pathologically proven prostate cancer.
Eur J Nucl Med.
, , , et al.
Are pelvic computed tomography, bone scan and pelvic lymphadenectomy necessary in the staging of prostatic cancer?
Br J Urol.
, , , et al.
Are serial bone scans useful for the follow-up of clinically localized, low to intermediate grade prostate cancer managed with watchful observation alone?
, , , et al.
Radionuclide bone scintigraphy in patients with biochemical recurrence after radical prostatectomy: when is it indicated?
, , , et al.
Distant metastasis after radical prostatectomy in patients without an elevated serum prostate specific antigen level.
, , , et al.
The incidence of prostate cancer progression with undetectable serum prostate specific antigen in a series of 394 radical prostatectomies.
, , .
Tibia metastasis without prostate specific antigen (PSA) increase following radical vesiculo-prostatectomy.
Int Urol Nephrol.
, , , et al.
M1 prostate cancer with a serum level of prostate-specific antigen less than 10 ng/mL.
Int J Urol.
, , , et al.
Limited value of bone scintigraphy and computed tomography in assessing biochemical failure after radical prostatectomy.
, , , et al.
Low PSA metastatic androgen-independent prostate cancer.
, , , et al.
Small cell and anaplastic prostate cancer: correlation between CT findings and prostate-specific antigen level.
, , , et al.
Seminal vesicle-secreted proteins and their reactions during gelation and liquefaction of human semen.
J Clin Invest.
, , , et al.
Decreased concentrations of prostate-specific antigen and human glandular kallikrein 2 in malignant versus nonmalignant prostatic tissue.
Undifferentiated carcinoma of the prostate with small cell features: immunohistochemical subtyping and reflections on histogenesis.
, , .
Stem-cell origin of metastasis and heterogeneity in solid tumors.
, , , et al.
Progression from adenocarcinoma to small cell carcinoma of the prostate with normalization of prostate-specific antigen (PSA) levels.
Scand J Urol Nephrol.
, , , et al.
Prostate carcinoma with testicular or penile metastases clinical, pathologic, and immunohistochemical features.