Skip Navigation

Japanese Journal of Clinical Oncology 2007 37(6):446-451; doi:10.1093/jjco/hym043
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Request Permissions
Google Scholar
Right arrow Articles by Shimizu, F.
Right arrow Articles by Fujime, M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Shimizu, F.
Right arrow Articles by Fujime, M.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?


© 2007 Foundation for Promotion of Cancer Research

Efficiency of Ultrasensitive Prostate-specific Antigen Assay in Diagnosing Biochemical Failure After Radical Prostatectomy

Fumitaka Shimizu1,, Shiro Tanaka2, Yutaka Matsuyama2, Takashi Tominaga3, Yasuo Ohashi2 and Makoto Fujime1

1 Department of Urology, Juntendo University, Tokyo
2 Department of Biostatistics/Epidemiology and Preventive Health Sciences, University of Tokyo, Tokyo
3 Department of Urology, Mitsui Memorial Hospital, Tokyo, Japan

For reprints and all correspondence: Fumitaka Shimizu, Department of Urology, Juntendo University, 3-1-3 Hongo, Bunkyo-ku, Tokyo 113-8431, Japan. E-mail: fshimizu-jua{at}umin.ac.jp

Received August 9, 2006; accepted February 1, 2007


   Abstract
TOP
 Abstract
INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 Conflict of interest statement
 References
 
Background: Ultrasensitive prostate-specific antigen (PSA) is a significant serum biomarker for identifying the PSA nadir and early biochemical failure after radical prostatectomy (RP). We assessed the efficiency of ultrasensitive PSA assay in the follow-up after RP.

Methods: We generated longitudinal ultrasensitive PSA data using a computer program assuming that patients experienced biochemical failure after RP. The simulation experiments, based on several different scenarios, were performed to assess the sensitivity and specificity in the diagnosis of biochemical failure using ultrasensitive PSA values and to estimate the lead time, which is the time advantage of detecting positivity for biochemical failure using the ultrasensitive PSA values compared with conventional PSA assay. We validated the sensitivity, specificity and lead time using actual follow-up data of 182 patients receiving RP.

Results: It was suggested that the sensitivity obtained from the actual data was more similar to that obtained using ultrasensitive PSA with an exponential increase than with a linear increase in the simulation experiments. Diagnosing biochemical failure based on two consecutive increases in the ultrasensitive PSA values was not recommended. Of non-biochemical failure patients, 9.4% showed four consecutive increases in their ultrasensitive PSA values. Average lead time in the actual data was 11.2 months (SD: 10.1).

Conclusions: For an accurate diagnosis of biochemical failure, our findings suggest the importance of a certain duration of follow-up and exclusion of false-positive results afterwards.

Key Words: prostatic neoplasms • prostatectomy • recurrence • prostate-specific antigen • computer simulation


   INTRODUCTION
TOP
 Abstract
 INTRODUCTION
METHODS
 RESULTS
 DISCUSSION
 Conflict of interest statement
 References
 
Recently prostate cancer has been increasingly detected earlier, which is attributable to the widespread use of prostate-specific antigen (PSA) screening. In addition, there has been a downward stage migration in patients diagnosed with prostate cancer (1). Previous studies reported that 15–39% of patients experienced biochemical failure after radical prostatectomy (RP) (2,3). Freeland et al. reported that patients with post-operative PSA values greater than 0.2 ng/ml are at a very high risk of developing an additional increase in PSA (4), whereas Stephenson et al. reported that biochemical failure defined as a PSA value of 0.4 ng/ml followed by another increase best predicts the development of distant metastasis (5). Some institutions have used the newly developed ultrasensitive PSA assay after RP instead of the conventional one. The third-generation ultrasensitive PSA assay is capable of measuring PSA levels below 0.02 ng/ml (6). The Immulite third-generation PSA assay (Diagnostic Products Corp., Los Angeles, California) is an assay with the lower limit of detection of 0.003 ng/ml (7). Several investigators reported that the ultrasensitive PSA nadir predicted the risk of early biochemical failure (8–10). It is reported that time to ultrasensitive PSA nadir also predicts early biochemical failure (10). However, the definition of biochemical failure using ultrasensitive PSA values after RP has not been standardized. Yu et al. (8) defined biochemical failure using ultrasensitive PSA values after RP as: (1) two or more consecutive increases in ultrasensitive PSA values that resulted in at least doubling of the initial ultrasensitive PSA value, (2) any increases that resulted in ultrasensitive PSA values greater than 0.1 ng/ml, or (3) a 10-fold increase in ultrasensitive PSA values between two measurements. However, this definition has not been widely accepted yet. At present, only consecutive increases in ultrasensitive PSA values are recommended to diagnose biochemical failure (11). Recently, salvage radiotherapy has been reported to be initiated for patients with lower ultrasensitive PSA values (12). However, it is not known whether or not these treatments have contributed to survival.

As the first step to avoid overtreatment in the post-operative follow-up, it is important to grasp the sensitivity, specificity and time advantage of ultrasensitive PSA assay in diagnosing biochemical failure. We tried to predict these outcomes by simulation experiments on computer program and validate the results using actual clinical data.


   METHODS
TOP
 Abstract
 INTRODUCTION
 METHODS
RESULTS
 DISCUSSION
 Conflict of interest statement
 References
 
Data Generation
Using random numbers, we generated the longitudinal ultrasensitive PSA data assuming that the patients experienced biochemical failure after RP. The exponential or linear increases in ultrasensitive PSA values are assumed in the following formula (13).


Formula 043UM1

where ß0(ß'0) represents an intercept, ß1(ß'1) represents a regression coefficient, time represents the duration of follow-up after the ultrasensitive PSA nadir and {epsilon}({epsilon}') represents an error term.

An error term {epsilon} in model (a) was assumed to be normally distributed with mean 0 and variance {sigma}2, i.e. {epsilon} ~ N(0, {sigma}2). In order to make the variance of PSA in model (a) and that of PSA' in model (b) equal, an error term ({epsilon}') in model (b) was defined as randomly generated PSA in model (a) minus the expected PSA in model (a). When time = 0, the intercept parameter exp(ß0) in model (a) or ß'0 in model (b) was set to be 0.003 ng/ml as the value of post-operative ultrasensitive PSA nadir. The value of the standard deviation (SD), {sigma}, was chosen in terms of the coefficient of variation (CV), the value of which was obtained from the reproducibility test in the brochure and previous studies (7,14). CV was obtained by measuring the same sample five to ten times with different kinds of ultrasensitive PSA kits. We fixed the prostate-specific antigen doubling time (PSADT) as the speed of tumor recurrence, and then the longitudinal ultrasensitive PSA data were randomly generated according to model (a) or (b). A computer program was written to generate random numbers using SAS Ver.9.1.3.

Settings of Simulations
We performed two simulation experiments to assess the sensitivity, specificity and lead time. The scenarios of each simulation experiment are shown in Tables 1 and 2. Longitudinal ultrasensitive PSA data of 1000 patients for the combinations of each scenario were generated using random numbers. In the first simulation experiment (Simulation 1), the value of PSADT in the biochemical failure group was set at 3, 6, 9 or 12 months, while that of patients in the control group, who had no recurrence, was set at 120 months. In the exponential increase, ß1 of 3, 6, 9, 12, or 120 months of PSADT was set at 2.3 x 10–1, 1.2 x 10–1, 7.7 x 10–2, 5.5 x 10–3, or 5.5 x 10–4 respectively. In the linear increase, ß'0 of 3, 6, 9, 12, or 120 months of PSADT was set at 1.0 x 10–3, 5.0 x 10–4, 3.3 x 10–4, 2.5 x 10–4, or 2.5 x 10–5 respectively. Positivity for biochemical failure was defined as two, three, or four consecutive increases in ultrasensitive PSA values. The duration of follow-up was set at 12 or 24 months, and the interval of ultrasensitive PSA measurements was set at 3 months. We chose the SD of {sigma} = 0.075 (best case) or {sigma} = 0.15 (worst case).


View this table:
[in this window]
[in a new window]

  		Table 1. Scenarios in Simulation 1

 

View this table:
[in this window]
[in a new window]

  		Table 2. Scenarios in Simulation 2

 
In the second simulation experiment (Simulation 2), all the data were generated as the biochemical failure group. Positivity for biochemical failure was defined as three consecutive increases in ultrasensitive PSA values. Biochemical failure was defined as having a PSA value greater than 0.2 ng/ml, which was the lower limit of detection of the Hybritech Tandem-R assay ({sigma} = 0.075). The duration of follow-up was defined as the time to positivity identified using an ultrasensitive PSA assay. The interval of ultrasensitive PSA measurements and SD ({sigma}) were set at the same values as those of model (a) in Simulation 1.

Outcome Measure for Simulations
In Simulation 1, the sensitivity and specificity were assessed in each scenario. The sensitivity was defined as the probability of enabling the identification of positivity in the biochemical failure group during the duration of follow-up. The specificity was defined as the probability of negativity being identified in the control group during the duration of follow-up.

In Simulation 2, we calculated the average lead time, which is the time difference between positivity identified using the ultrasensitive PSA assay and biochemical failure identified using the conventional PSA assay. A graphical representation of lead time is shown in Figure 1.


Figure 1
View larger version (8K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 1. A graphical representation of the lead time in Simulation 2. PSA, prostate-specific antigen.

 
Validation of Simulation Results Using Actual Data
Between January 1999 and December 2004, 302 men with a clinically localized prostate cancer underwent radical retropubic prostatectomy at two institutions. Of these patients, 120 were excluded from the validation data set owing to insufficient follow-up in 48 men, administration of adjuvant therapy in 40 men, follow-up using a conventional assay in 24 men and the ultrasensitive PSA nadir being 0.05 ng/ml or more in 8 men. The remaining 182 patients were post-operatively followed every one to three months using an ultrasensitive assay and were included in the current analysis. Of these patients, 23 (12.6%) experienced biochemical failure. Biochemical failure was defined as PSA being greater than 0.2 ng/ml. Positivity for biochemical failure was defined as two, three or four consecutive increases and we calculated the sensitivity and specificity in biochemical failure. We also calculated the time difference between the time to positivity, which was defined as three consecutive increases and the time to a value greater than 0.2 ng/ml.


   RESULTS
TOP
 Abstract
 INTRODUCTION
 METHODS
 RESULTS
DISCUSSION
 Conflict of interest statement
 References
 
We assessed the sensitivity, specificity, and lead time using the longitudinal ultrasensitive PSA data generated with random numbers. The results of Simulation 1 are shown in Tables 3 and 4.


View this table:
[in this window]
[in a new window]

  		Table 3. The sensitivity and specificity of biochemical failure using an ultrasensitive PSA during a 12 months follow-up in Simulation 1

 

View this table:
[in this window]
[in a new window]

  		Table 4. The sensitivity and specificity of biochemical failure using an ultrasensitive PSA during a 24-month follow-up in Simulation 1

 
The sensitivity differed remarkably between the exponential and linear increases in ultrasensitive PSA values. In the exponential increases, as the duration of follow-up became longer, the sensitivity increased in all scenarios. However, the sensitivity decreased with the increasing number of consecutive increases. In the linear increases, when positivity for biochemical failure was defined as three or four consecutive increases, the sensitivity presented as a remarkably low value without being influenced by the duration of follow-up. In contrast to the sensitivity, the specificity did not differ remarkably between the exponential and linear increases in ultrasensitive PSA values. The results of specificity based on two consecutive increases presented low value without being influenced by the duration of follow-up. The sensitivity and specificity which were calculated using actual data are shown in Table 5. The sensitivity obtained from the assumption that ultrasensitive PSA increased exponentially in Simulation 1 was more similar to the sensitivity obtained from the actual data than that in which the linear increases were assumed. Of the biochemical failure group, 0%, 4.3% and 13.0% did not experience two, three and four consecutive increases, respectively. However, of the no recurrence group, 42.8%, 18.9% and 9.4 % experienced two, three and four consecutive increases, respectively. In Simulation 2 and the actual data, the lead time between positivity identified using the ultrasensitive PSA assay and biochemical failure identified using the conventional PSA assay is shown in Table 6. The average lead time was approximately 5 years when the value of PSADT was 12 months in the simulation data. The average of PSADT in patients who experienced biochemical failure in the actual data was 5.6 months. The average, minimum and maximum lead time in the actual data was 11.2, 0 and 38.0 months, respectively.


View this table:
[in this window]
[in a new window]

  		Table 5. The sensitivity and specificity of biochemical failure using an ultrasensitive PSA in actual data

 

View this table:
[in this window]
[in a new window]

  		Table 6. The lead time between positivity identified using an ultrasensitive PSA assay and biochemical failure identified using a conventional PSA assay in Simulation 2 and actual data

 

   DISCUSSION
TOP
 Abstract
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
Conflict of interest statement
 References
 
There has been an increase in the number of patients receiving salvage radiotherapy for biochemical failure after RP. In the past, these patients had been administered salvage radiotherapy after clinical local recurrence (15). Recently, Cheung et al. reported that earlier initiation of salvage radiotherapy after post-operative biochemical failure led to a good outcome (16). However, if biochemical failure is identified using low ultrasensitive PSA values, unnecessary salvage therapies may be administered to the false-positive patients. In addition, some of the patients whose life expectancy is not long may not even require salvage therapy. We hypothetically set up the biochemical failure group and assessed the efficiency of ultrasensitive PSA assay after RP through Monte Carlo simulation experiments.

Increases in conventional PSA values after RP have been reported to represent an exponential growth curve (3,17), whereas the significance of those in ultrasenstive PSA values has remained unproven until the present. The sensitivity of actual data more closely resembled the result from exponential increases rather than that from linear increases in biochemical failure after RP in Simulation 1. More specifically, it was suggested that PSADT assuming an exponential increase might be more suitable for predicting post-operative biochemical failure using an ultrasensitive PSA assay than PSA velocity assuming a linear increase. Next, using actual data, some patients experienced biochemical failure without identifying four consecutive increases in ultrasensitive PSA values which represented false-negative, whereas other patients were free of biochemical failure, even if four consecutive increases were confirmed, which represented false-positive. We need to reduce false-positivity in the post-operative follow-up, because the situation is different from screening. Therefore, it is required to extend the duration of follow-up and then increase the number of consecutive increase in ultrasensitive PSA values. Shinghal et al. reported a significant variation in ultrasensitive PSA values after RP without clinical progression during long-term follow-up (18). Djavan et al. reported that given the small amount of PSA produced and the lack of PSA progression, residual benign glands and periurethral glands might be the source of PSA (19). Accordingly, salvage therapies should be adapted carefully. Although using an ultrasensitive PSA assay is helpful for grasping post-operative PSA nadir as compared with conventional PSA, it may be difficult to define biochemical failure using consecutive increases in ultrasensitive PSA values.

In our data, ultrasensitive PSA assay enabled a time advantage of 11.2 months to be gained in the diagnosis of biochemical failure. It has been reported in previous studies that the lead time of biochemical failure identified using ultrasensitive PSA and conventional PSA assays ranged from 10 to 29 months (6,20). However, these values were averages and differences arising from the speed of tumor recurrence were not taken into consideration. Several investigators reported that the value of PSADT at biochemical failure after RP might be the surrogate endpoints for a prostate cancer-specific mortality or a predictor of biochemical outcome for salvage radiotherapy (21–24). We estimated the lead time by PSADT in Simulation 2. Pound et al. (3) reported that the median actuarial time to clinical metastasis was 8 years from the time of elevation of the PSA level after RP and that once men developed metastatic disease, the median actuarial time to death was 5 years. Using the ultrasensitive PSA assay, the time from positivity for biochemical failure after RP to death will be much longer. D'Amico et al. reported that patients with a post-operative PSADT of more than 12 months might not require salvage radiotherapy (25). The estimated lead time in Simulation 2 may help in making decisions on salvage therapies that appropriately correspond to life expectancy. However, we need to recognize the instability of the calculated PSADT arising from the imprecision of the ultrasensitive PSA assay compared with the conventional assay.

We admit a limitation of our study that we could not estimate the sensitivity and lead time using actual data stratified by PSADT, because there were few patients with biochemical failure who did not receive adjuvant therapy after RP.

In conclusion, although using an ultrasensitive PSA assay in post-operative monitoring may make early detection of a biochemical failure possible, the risk associated with the administration of unnecessary therapy will also increase. In order to diagnose the biochemical failure correctly, our findings suggest it is important to resort to a certain duration follow-up, and then introduce salvage therapies carefully.


   Conflict of interest statement
TOP
 Abstract
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 Conflict of interest statement
References
 
None declared.


   References
TOP
 Abstract
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 Conflict of interest statement
 References
 
1 Han M, Partin AW, Piantadosi S, Epstein JI, Walsh PC. Era specific biochemical recurrence-free survival following radical prostatectomy for clinically localized prostate cancer. J Urol (2001) 166:416–9.[CrossRef][Web of Science][Medline]

2 Han M, Partin AW, Zahurak M, Piantadosi S, Epstein JI, Walsh PC. Biochemical (prostate specific antigen) recurrence probability following radical prostatectomy for clinically localized prostate cancer. J Urol (2003) 169:517–23.[CrossRef][Web of Science][Medline]

3 Pound CR, Partin AW, Eisenberger MA, Chan DW, Pearson JD, Walsh PC. Natural history of progression after PSA elevation following radical prostatectomy. JAMA (1999) 281:1591–7.[Abstract/Free Full Text]

4 Freedland SJ, Sutter ME, Dorey F, Aronson WJ. Defining the ideal cutpoint for determining PSA recurrence after radical prostatectomy. Prostate-specific antigen. Urology (2003) 61:365–9.[CrossRef][Web of Science][Medline]

5 Stephenson AJ, Kattan MW, Eastham JA, Dotan ZA, Blanco FJ Jr, Lilja H, et al. Defining biochemical recurrence of prostate cancer after radical prostatectomy: a proposal for a standardized definition. J Clin Oncol (2006) 24:3973–8.[Abstract/Free Full Text]

6 Yu H, Diamandis EP, Prestigiacomo AF, Stamey TA. Ultrasensitive assay of prostate-specific antigen used for early detection of prostate cancer relapse and estimation of tumor-doubling time after radical prostatectomy. Clin Chem (1995) 41:430–4.[Abstract/Free Full Text]

7 Hayashi A. Clinical utility of the ultrasensitive prostate specific antigen assay ‘IMMULYZE HS-PSA. J Anal Bio-Sci (2004) 27:284–90.

8 Yu H, Diamandis EP, Wong PY, Nam R, Trachtenberg J. Detection of prostate cancer relapse with prostate specific antigen monitoring at levels of 0.001 to 0.1 microG/L. J Urol (1997) 157:913–8.[CrossRef][Web of Science][Medline]

9 Shen S, Lepor H, Yaffee R, Taneja SS. Ultrasensitive serum prostate specific antigen nadir accurately predicts the risk of early relapse after radical prostatectomy. J Urol (2005) 173:777–80.[CrossRef][Web of Science][Medline]

10 Nakamura M, Hasumi H, Miyoshi Y, Sugiura S, Fujinami K, Yao M, et al. Usefulness of ultrasensitive prostate-specific antigen assay for early detection of biochemical failure after radical prostatectomy. Int J Urol (2005) 12:1050–4.[CrossRef][Web of Science][Medline]

11 Japanese Urological Association and the Japanese Society of Pathology, editors. General Rule for Clinical and Pathological Studies on Prostate Cancer (2001) 3rd edn. Tokyo: Kanahara.

12 Terai A, Matsui Y, Yoshimura K, Arai Y, Dodo Y. Salvage radiotherapy for biochemical recurrence after radical prostatectomy. BJU Int (2005) 96:1009–13.[CrossRef][Web of Science][Medline]

13 Schmid HP, McNeal JE, Stamey TA. Observations on the doubling time of prostate cancer. The use of serial prostate-specific antigen in patients with untreated disease as a measure of increasing cancer volume. Cancer (1993) 71:2031–40.[CrossRef][Web of Science][Medline]

14 Chikahira Y, Doi F, Hasegawa E, Hosooka S, Sasagawa N, Matsuda N, et al. Ultrasensitive assay of prostate-specific antigen in radical prostatectomy using Lumipulse ‘PSA-N. J Anal Bio-Sci (2004) 27:279–83.

15 Ornstein DK, Colberg JW, Virgo KS, Chan D, Johnson ET, Oh J, et al. Evaluation and management of men whose radical prostatectomies failed: results of an international survey. Urology (1998) 52:1047–54.[CrossRef][Web of Science][Medline]

16 Cheung R, Kamat AM, de Crevoisier R, Allen PK, Lee AK, Tucker SL, et al. Outcome of salvage radiotherapy for biochemical failure after radical prostatectomy with or without hormonal therapy. Int J Radiat Oncol Biol Phys (2005) 63:134–40.[CrossRef][Web of Science][Medline]

17 Patel A, Dorey F, Franklin J, deKernion JB. Recurrence patterns after radical retropubic prostatectomy: clinical usefulness of prostate specific antigen doubling times and log slope prostate specific antigen. J Urol (1997) 158:1441–5.[CrossRef][Web of Science][Medline]

18 Shinghal R, Yemoto C, McNeal JE, Brooks JD. Biochemical recurrence without PSA progression characterizes a subset of patients after radical prostatectomy. Prostate-specific antigen. Urology (2003) 61:380–5.[CrossRef][Web of Science][Medline]

19 Djavan B, Sesterhann I, Hruby S, Susani M, Haitel A, Etemad M, et al. Benign prostatic glands in the surgical margin of radical retropubic prostatectomies: redefining PSA nadir. J Urol (2000) 163:A624.

20 Haese A, Huland E, Graefen M, Hammerer P, Noldus J, Huland H. Ultrasensitive detection of prostate specific antigen in the followup of 422 patients after radical prostatectomy. J Urol (1999) 161:1206–11.[CrossRef][Web of Science][Medline]

21 Freedland SJ, Humphreys EB, Mangold LA, Eisenberger M, Dorey FJ, Walsh PC, et al. Risk of prostate cancer-specific mortality following biochemical recurrence after radical prostatectomy. JAMA (2005) 294:433–9.[Abstract/Free Full Text]

22 Ward JF, Zincke H, Bergstralh EJ, Slezak JM, Blute ML. Prostate specific antigen doubling time subsequent to radical prostatectomy as a prognosticator of outcome following salvage radiotherapy. J Urol (2004) 172:2244–8.[CrossRef][Web of Science][Medline]

23 Leventis AK, Shariat SF, Kattan MW, Butler EB, Wheeler TM, Slawin KM. Prediction of response to salvage radiation therapy in patients with prostate cancer recurrence after radical prostatectomy. J Clin Oncol (2001) 19:1030–9.[Abstract/Free Full Text]

24 Stephenson AJ, Shariat SF, Zelefsky MJ, Kattan MW, Butler EB, Teh BS, et al. Salvage radiotherapy for recurrent prostate cancer after radical prostatectomy. JAMA (2004) 291:1325–32.[Abstract/Free Full Text]

25 D'Amico AV, Chen MH, Roehl KA, Catalona WJ. Identifying patients at risk for significant versus clinically insignificant postoperative prostate-specific antigen failure. J Clin Oncol (2005) 23:4975–9.[Abstract/Free Full Text]


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?



This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Request Permissions
Google Scholar
Right arrow Articles by Shimizu, F.
Right arrow Articles by Fujime, M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Shimizu, F.
Right arrow Articles by Fujime, M.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?