Clinical evaluation of a novel method for the measurement of prostate-specific antigen, AccuPSATM, as a predictor of 5-year biochemical recurrence-free survival after radical prostatectomy: results of a pilot study


Herbert Lepor, Department of Urology, NYU School of Medicine, 150 E. 32nd Street, New York, NY 10016, USA. e-mail:


Study Type – Diagnostic (validating cohort)

Level of Evidence 1b

What's known on the subject? and What does the study add?

Nadir Ultrasensitive PSA levels has some value for predicting BCR following RD.

AccuPSA assays lower limit of PSA quantification of <0.01 pg/ml greatly enhances sensitivity and specificity of nadir PSA to predict BCR following RP. Our pilot study shows an AccuPSA of 3 pg/ml has a sensitory and specificity of 100% and 75% respectively for predicting 5 year BCR following RP.


  • • To conduct a proof of concept study to evaluate a novel digital single molecule immunoassay (AccuPSATM) that detects prostate-specific antigen (PSA) a thousandfold more sensitively than current PSA detection methods.
  • • To determine the ability of the AccuPSATM assay to predict 5-year biochemical recurrence (BCR)-free survival after radical prostatectomy (RP).


  • • A total of 31 frozen serum specimens were obtained from specimen logs maintained at New York University Langone Medical Center and the Johns Hopkins University School of Medicine on men who had undergone RP. Those men without evidence of BCR had a minimum of 5 years' PSA follow-up.
  • • In all cases, preoperative and pathological information were available, as was a serum specimen 3–6 months after RP, with a PSA level of <0.1 ng/mL measured by conventional PSA methods at the time of serum collection.
  • • Specimens were tested using the AccuPSATM method.
  • • A Cox proportional hazard model and Kaplan–Meier analysis were used to determine whether AccuPSATM predicted the risk of BCR.


  • • Overall, 11/31 (35.5%) men developed BCR.
  • • Mean AccuPSATM nadir levels were significantly different (P < 0.001) between the non-BCR group (2.27 pg/mL) and the BCR group (46.99 pg/mL).
  • • Using a multivariate Cox proportional hazard model, AccuPSATM nadir level was a significant predictor of BCR-free survival (P < 0.01).
  • • Kaplan–Meier analysis of up to 5 years follow-up showed that 100% of men with AccuPSATM nadir values <3 pg/mL did not develop BCR, whereas 62.5% of men with values >3 pg/mL developed BCR (P= 0.00024).
  • • The sensitivity, specificity, positive predictive value and negative predictive value of the AccuPSATM method was 100%, 75%, 69% and 100%, respectively.


  • • AccuPSATM assay predicts 5-year BCR- free survival after RP.
  • • Identifying a reliable predictor of BCR soon after RP has important implications for frequency of PSA testing, selection of candidates for adjuvant therapy, and reassuring a large subset of men that they are not at risk of recurrence.
  • • Larger studies are needed to validate these findings.

biochemical recurrence


radical prostatectomy


open radical retropubic prostatectomy


positive predictive value


negative predictive value


Prostate-specific antigen is a serine protease produced almost exclusively by the epithelial elements of the prostate [1]. Serum assays detecting PSA were first approved by the US Food and Drug Administration (FDA) in 1986 for monitoring prostate cancer after treatment [2]. It was not until 1994 that an assay measuring serum PSA was approved by the FDA for the early detection of prostate cancer in combination with DRE [3,4]. A major limitation of the use of PSA for screening/early detection is its lack of specificity, owing to its production by benign as well as malignant prostatic epithelium.

Theoretically, PSA should be a useful tool for monitoring the effectiveness of radical prostatectomy (RP) to eradicate the disease since all of the prostate tissue should be removed in this procedure. Residual local spread or systemic metastases of prostate cancer after RP is manifested as a measurable PSA level, which increases over time depending on the extent of disease [5]. A recent study provides compelling evidence that benign residual prostate tissue is a very rare cause of measureable PSA after RP [6].

There are several potential advantages of predicting residual disease after RP based on the nadir or initial post-prostatectomy PSA measurement. First, those men who are destined not to develop disease recurrence may be reassured soon after RP, thereby alleviating anxiety and the need for extended monitoring. Second, those destined to develop recurrent disease may be offered adjuvant treatment if clinically indicated at an earlier time point.

The detection limit for the first approved PSA assays was 0.1 ng/mL [7]. In the modern era, the overwhelming majority of post-prostatectomy PSA levels are undetectable at a value <0.1 ng/mL, which means the first or nadir hypersensitive PSA assay is of limited value for assessing residual disease [8]. Ultrasensitive PSA assays were later commercialized with detection limits of 0.01 ng/mL [9,10]. Several studies showed that ultrasensitive PSA nadir levels between 0.01 and 0.04 ng/mL were independent predictors of biochemical recurrence (BCR) after RP [11–14]. While the ultrasensitive PSA nadir level identified a subset of men with a higher risk of BCR, this level lacked sensitivity; however, the results suggested nadir PSA was useful, although it required better separation of the BCR and non-BCR groups. The fact that the majority of men who ultimately developed BCR had a nadir PSA <0.01 ng/mL provides the rationale for determining if nadir PSA assays with even lower limits of detection will enhance prediction of and the timing of BCR.

AccuPSATM employs a novel single molecule assay that has a lower limit of PSA quantification of <0.01 pg/mL. This is a thousandfold lower than conventional ultrasensitive PSA assays [15,16]. The primary objective of this proof of concept study was to determine the usefulness of the nadir post-prostatectomy AccuPSATM method of detecting PSA for predicting 5-year BCR-free survival after RP.



A total of 31 frozen serum specimens were obtained from specimen logs maintained at New York University Langone Medical Center and the Johns Hopkins University School of Medicine on men who had undergone open radical retropubic prostatectomy (ORRP) with a minimum of 5-years' PSA follow-up for those without evidence of BCR. For all men, a serum specimen was obtained 3–6 months after ORRP. This is referred to as the nadir sample. All specimens were required to have had a PSA level of <0.1 ng/mL measured by conventional PSA detection methods at the time of serum collection for entry into the study. Men with evidence of nodal or distant metastases at the time of surgery were excluded from the study. No subjects received neo-adjuvant or adjuvant hormonal or radiation treatment.

Baseline demographic information and data on preoperative serum PSA level, clinical stage, Gleason score of the prostate biopsy, pathological stage and Gleason score, surgical margin status, PSA nadir and subsequent PSA levels, date of BCR and date of any secondary prostate cancer treatment were maintained prospectively as part of longitudinal prospective Institutional Review Board-approved databases and specimen biorepositories at the respective institutions. BCR was defined as two consecutive PSA measurements >0.2 ng/mL after the initial collected sample, or secondary treatment for progressively rising serum PSA level.

All serum samples were kept frozen at −70° C or colder from the time of initial collection. Specimens were shipped on dry ice for AccuPSATM testing at Quanterix Corporation (Cambridge, MA, USA).


The AccuPSATM method is a single molecule digital ELISA with sub-femtomolar detection limits of serum PSA. The technology has been previously reported [15,16]. Briefly described, this novel technology detects single protein molecules in blood by capturing the proteins on microscopic beads decorated with specific antibodies and labelling the immunocomplexes with a reporter capable of generating a fluorescent product. After isolating the beads in 50-femtolitre reaction chambers designed to hold a single bead, fluorescence imaging detects the single protein molecules. The AccuPSATM method has been shown to provide linear response over ≈ four logs of concentration ([PSA] from 8 fg/mL to 100 pg/mL) and extends the dynamic range of ELISA from picomolar levels down to sub-femtomolar levels in a single measurement [15]. The analytical performance of the AccuPSATM method has been shown (Wilson DH et al., unpublished data).


A Cox proportional hazard model was performed to determine whether AccuPSATM level predicted the risk of BCR. The covariates that were entered into the regression equation were age at radical prostatectomy, PSA value before surgery (ng/mL), biopsy Gleason Score, clinical Stage (T1,T2), nadir value of AccuPSATM (pg/mL), pathological Gleason Score, pathological stage (pT2, pT3) and margin status (negative or positive). Forward elimination using the likelihood ratio test was employed. A P value <0.05 was considered to indicate statistical significance.

A bootstrapped 95% CI for the nadir AccuPSATM value in the non-BCR group was used to determine the AccuPSATM cut-off value that defined two risk groups.

Kaplan–Meier survival curves, stratified by the bifurcated AccuPSATM nadir value were used to examine time to 5-year BCR after ORRP. A Student's t-test was used to determine the difference between mean nadir AccuPSATM values for those men who had BCR within 5 years and those men who did not. All analyses were performed using SPSS, Version 18 (IBM, NY, USA).


The recorded characteristics of the 31 men who underwent ORRP and fulfilled the study criteria are shown in Table 1. Overall, 11 (35.5%) developed BCR. The relevant characteristics are compared between the BCR and non-BCR groups. Age at ORRP and race were similar between the groups. The group of men who developed BCR, developed it within a mean of 2.1 years after ORRP and had a higher pre-surgical PSA level, clinical and pathological stage, Gleason score and grade than the non-BCR group of men. Margin status was similar between the groups.

Table 1.  Characteristics of men undergoing ORRP in the BCR and non-BCR groups
 Non-BCR group(n= 20)BCR group(n= 11)
Mean age at RP, years61.561.0
Mean time to event, years5.02.1
Mean pre-surgical PSA, ng/mL6.410.3
Race, n (%)  
 Caucasian19 (95)10 (91)
 African-American 1 ( 9)
 Unknown1 (5) 
Gleason score, n (%)  
 51 (5) 
 613 (65)4 (36)
 73 (15)7 (64)
 83 (15) 
Clinical stage, n (%)  
 T1c15 (75 )5 (45)
 T2a4 ( 20)3 (27 )
 T2b1 (5)3 (27)
Gleason grade, n (%)  
 Missing 1 (9)
 2 + 31 (5) 
 3 + 313 (65)4 (36)
 3 + 41 (5)3 (27)
 4 + 32 (10)3 (27)
 4 + 43 (15) 
Pathology stage, n (%)  
 pT2a4 (20)1 (9)
 pT2b6 (30) 
 pT2c7 (35)1 (9)
 pT3a1 (5)5 (45)
 pT3b 3 (27)
 Missing2 (10)1 (9)
Margin status, n (%)  
 Negative19 (95)10 (91)
 Positive1 (5)1 (9)
Pathology: Gleason grade, n (%)  
 Missing1 (5) 
 3 + 21 (5)1 (9)
 3 + 310 (50)1 (9)
 3 + 45 (25)5 (45)
 4 + 3 2 (18)
 4 + 41 (5) 
 4 + 51 (5)2 (18)
 5 + 41 (5) 

The distribution of the nadir AccuPSATM levels and the nadir AccuPSATM statistics for the BCR and non-BCR groups are shown in Fig. 1 and Table 2, respectively. The mean AccuPSATM levels in the non-BCR and BCR groups were 2.27 pg/mL and 46.99 pg/mL, respectively (P < 0.001). Although PSA <0.1 ng/mL was one of the study inclusion criteria, nadir values for two men exceeded this value as measured by AccuPSATM and were not excluded from the study. Differences in standardization and high variability at the detection limit of the conventional PSA methods could account for this discrepancy [17].

Figure 1.

Dot plot of Ln AccuPSATM nadir (pg/mL) for non-BCR and BCR groups.

Table 2.  Nadir AccuPSATM levels stratified by BCR and non-BCR groups
 Non-BCR, pg/mLBCR, pg/mL

Only pathological stage and AccuPSATM were predictors of BCR. Cox-multivariate regression analysis was performed and nadir AccuPSATM remained an independent predictor of BCR (Table 3). The parameter estimate for AccuPSATM (B = 0.014) indicates that at any given time the risk of recurring will increase by 1.01% for every 1 pg/mL increase in the nadir PSA level.

Table 3.  Cox regression analysis of BCR
 Variables in the equation
BseWalddfPExp. (B)
  1. B, regression coefficient CLN (Hazard Ratio); SE, standard error; Wald, Wald test statistic (square of the ratio BISE); P, p value; Exp. (B), estimate of the risk ratio for every unit of change in B.

Step 1Nadir PSA0.0190.0077.84210.0051.019
Step 2Pathological stage (pT2, pT3)2.0250.8415.80110.0167.579
Nadir PSA0.0140.0074.41710.0361.014

A bootstrapped 95% CI for the nadir AccuPSATM level had an upper limit of 2.9 pg/mL. The value of 3.0 pg/mL was used as a threshold for defining the two risk groups (high vs low) for BCR. The Kaplan–Meier survival curves for the risk groups defined by the bifurcated nadir value is shown in Fig. 2. P= 0.00024 for the difference in BCR-free survival between the two plots.

Figure 2.

Kaplan–Meier time-to-BCR curves.

The derivation of the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for the nadir AccuPSATM threshold of 3 pg/mL for predicting BCR within 5 years is shown in Table 4. The sensitivity, specificity, PPV and NPV were 100%, 75%, 69% and 100%, respectively.

Table 4.  Sensitivity, specificity, PPV and NPV of AccuPSATM at a threshold of 3 pg/mL
 BCR in <5 yearsNo BCR in ≥5 yearsPPV/NPV
AccuPSATM≥3 pg/mL11569% PPV
AccuPSATM <3 pg/mL015100% NPV


One of the primary attributes of a serum PSA assay is the ability to monitor prostate cancer disease progression after definitive therapy. The rate of PSA change has been shown to be a prognostic indicator of disease-specific survival after RP [18], radiation therapy [19] or androgen deprivation therapy [20], therefore, a persistently rising PSA level is indicative of disease progression after any treatment intervention.

The advantages of early prediction of BCR after RP include the opportunity to provide reassurance to those unlikely to have BCR and the opportunity to initiate timely adjuvant treatment for those highly likely to have BCR. Current predictive models do not reliably identify the groups at risk of BCR after RP [21]. For example, a Southwest Oncology Group study [22] randomized men at ‘high risk’ for disease progression after RP to either an ‘adjuvant radiation therapy’ group or a ‘no treatment’ group. While the group undergoing adjuvant radiation therapy had a significantly lower risk of disease progression, over half of the men in the untreated group never developed BCR, which indicated that over half of the men received unnecessary RT. The study also failed to show whether adjuvant radiation therapy is superior to salvage radiation therapy. In this setting, men must balance the reassurance of unproven benefits of early intervention against the complications associated with unnecessary intervention. It is apparent that we need better early predictors of BCR after RP.

The ideal scenario would be the opportunity to reliably predict BCR based on the nadir PSA level. The clinical usefulness of nadir PSA levels for predicting BCR after RP is highly controversial. As the ability to assay lower levels of PSA is developed, the specificity of the measurable PSA for cancer recurrence poses a clinical dilemma. The limited specificity of very low levels of PSA may be attributable to benign prostatic tissue, cross-reactivity with PSA-like molecules or production of PSA from the periurethral glands. A recent study by Godoy et al. [6] suggests that benign tissue is an extremely unlikely cause of measureable PSA (PSA level 0.04–0.15 ng/ml) or BCR. With a mean follow-up of 36 months, men with a very low risk of BCR based on clinical and pathological factors developed a measurable PSA or BCR in 0.6% and 0.3% of cases, respectively.

Ellis et al. [9] were some of the first investigators to evaluate the level of non-prostate cancer PSA in a cohort of men undergoing cystoprostatectomy which would represent the residual disease detection limit. Using the Hybritech chemiluminescent assay, the lower limit of detection, biological limit of detection and the residual disease detection limit of the PSA assay were 3.9 pg/mL, 8.0 pg/mL, and 30 pg/mL, respectively. Using this assay, they concluded that the ultrasensitive PSA provided a mean lead time for detecting BCR of 12.7–22.5 months compared with the conventional hypersensitive PSA assays widely used at the time in clinical practice. Yu et al. [23] also reported that an ultrasensitive PSA assay detecting concentrations <0.001 ng/mL decreased the time taken to identify prostate cancer relapse. In both studies, the endpoint was cancer relapse defined as a PSA >0.1 ng/mL and not BCR with PSA >0.2 ng/mL. The rate of PSA change was felt to be the best indicator of cancer relapse. Neither of the studies commented on the ability of nadir ultrasensitive PSA to predict BCR.

Shen et al. [13] were the first to examine the ability of ultrasensitive PSA to predict BCR in 545 men who underwent ORRP by a single surgeon [13]. Nadir PSA was defined using the Immulite® third generation PSA assay with a lower limit of detection of 0.01 ng/mL. The mean follow-up was 3.1 year and 9.9% of the group developed BCR, which was defined as two consecutive increasing post-nadir PSA levels and an absolute PSA level >0.1 ng/mL. Using a multivariate logistic regression analysis, nadir PSA was an independent predictor of BCR. The risk of BCR in men with a nadir PSA <0.01, 0.01, 0.02, and >0.04 ng/mL was 4.0%, 12.0%, 15.8% and 89.3%, respectively. While a nadir PSA >0.04 ng/mL indicates a poor prognosis, only 5% of men had a PSA this high.

Nakamura et al. [24] examined nadir ultrasensitive PSA level as a predictor of BCR using the Immulite® third generation PSA assay study, BCR was defined as a PSA >0.2 ng/mL. Of the 46 evaluable men, 15(32.6%) developed BCR. The mean follow-up was only 33 months which is too short to reliably identify all men who will develop a BCR. The mean nadir ultrasensitive PSA in the recurrence and non-recurrence group was 0.05 and 0.007 ng/mL, respectively (P < 0.001). The optimum sensitivity and specificity for predicting BCR was observed for a threshold ultrasensitive nadir PSA of 0.01 ng/mL with a sensitivity and specificity of 100% and 68.2%, respectively.

Eisenberg et al. [11] reported on nadir ultrasensitive PSA as a predictor of 5-year BCR. Of the 1674 men undergoing RP between 1996 and 2006, only 525 had a nadir ultrasensitive PSA and did not undergo adjuvant intervention. Since a uniform ultrasensitive PSA assay was not used, a PSA <0.05 ng/mL was considered undetectable. BCR was defined as PSA >0.2 ng/mL. BCR-free survival was 86% and 67%, respectively for undetectable and detectable PSA (P < 0.01). Nadir ultrasensitive PSA was an independent predictor of BCR. Only 13% of men had a measurable nadir PSA.

The prognostic significance of nadir PSA was more recently examined by Hong et al. [25] with an ultrasensitive radioimmunoassay capable of PSA detection of 0.001 ng/mL. This retrospective study reported on 384 men followed for a median of 39.4 months after RP. The men were stratified into four PSA nadir groups corresponding to <0.001 ng/mL (undetectable, group 1), 0.001 ≤ and <0.02 ng/mL (group 2), 0.02 ≤ and <0.05 ng/mL (group 3), and ≥0.05 ng/mL (group 4). Approximately one half the men exhibited undetectable nadir PSA. BCR-free survivals were 95.5%, 79.8%, 78.0% and 41.5% for subjects in groups 1, 2, 3 and 4, respectively and statistical differences were observed among the groups (P < 0.001). The study further shows that undetectable sensitive PSA nadir (0.001 ng/mL) is an independent predictor of BCR-free survival. Furthermore, the BCR rate of groups 1 and 2 suggest an ultra-low risk group in the undetectable realm below third-generation assay measurement limits.

Thaxton et al. [26] reported on a nanoparticle-based bio-barcode assay that detects PSA levels at 330 fg/mL which is >300 times lower than conventional assays. They recently reported on 18 men who had undergone RP. The small cohort was separated into two categories based on follow-up PSA levels and not nadir PSA. In fact, the nadir PSA on inspection of the data is of no predictive value. The small sample size, and presentation of individual patient plots with no summary data, limits the ability to predict the usefulness of this assay.

The present study examines the ability of AccuPSATM to predict 5-year BCR in men undergoing ORRP. This novel assay has been shown to detect PSA levels in men after RP as low as 14 fg/mL which is lowest measureable level reported to date.

The present study was designed to determine if a single nadir ultrasensitive PSA assay would predict BCR after RP. The nadir PSA level was significantly higher in the group who developed BCR. The twentyfold greater AccuPSATM in the group who developed BCR exceeds the tenfold difference observed by Nakamura et al. [24] using the Immulite® ultrasensitive PSA assay. Nadir AccuPSATM was an independent predictor of BCR. The AccuPSATM threshold of 3.0 pg/mL provided the best separation between the BCR and non-BCR groups. We recognize that a larger cohort will be required to determine the optimum threshold for AccuPSATM level.

A difference in mean AccuPSATM levels does not ensure clinical relevance unless there is limited overlap of the PSA levels between the recurrence and non-recurrence groups. The distribution of AccuPSATM levels observed in the present proof of concept study is encouraging.

There are two scenarios where the AccuPSATM level could affect post-prostatectomy management. First is the ability to reassure an individual that he truly has a low risk of disease recurrence which is reflected in the NPV. In the present study, the NPV of a AccuPSATM <3 pg/mL was 100%. If future studies confirm this observation then a large cohort of men will be reassured that their disease will not recur and the frequency of PSA testing can be decreased. Second is the ability to identify men earlier for adjuvant treatment which is reflected in the specificity of the assay. In the present study the specificity of AccuPSATM was 75%. If men with a nadir AccuPSATM underwent adjuvant radiation therapy then 25% of these men would undergo unnecessary treatment. While this is of concern, the specificity is superior to the current criteria for adjuvant RT recommended by Thompson et al. [22].

The major strength of the present study is that all men had appropriately stored frozen serum for a nadir AccuPSATM determination. The 5-year follow-up time interval we feel is sufficient to identify most cases of clinically significant BCR. Another strength of the study is that BCR was defined uniformly at both sites. The limitation of the study is the relatively small sample size. The samples do not represent a consecutive cohort of men undergoing RP. An effort was made to enrich the sample with men with BCR. This is reflected in the 35.5% rate of BCR. The encouraging results should be interpreted with cautious enthusiasm as we await an expanded multicentre clinical study.


This work was supported by Quanterix Corp (Cambridge, MA, USA) and Twin Lights Bioscience, Inc (Cambridge, MA, USA). This work was supported in part by Award Number R43CA133987 from the National Cancer Institute.


None declared.