Satoru Hayasaka

Controlling the familywise error rate in functional neuroimaging: a comparative review

Functional neuroimaging data embodies a massive multiple testing problem, where 100 000 correlated test statistics must be assessed. The familywise error rate, the chance of any false positives is the standard measure of Type I errors in multiple testing. In this paper we review and evaluate three approaches to thresholding images of test statistics: Bonferroni, random field and the permutation test. Owing to recent developments, improved Bonferroni procedures, such as Hochberg’s methods, are now applicable to dependent data. Continuous random field methods use the smoothness of the image to adapt to the severity of the multiple testing problem. Also, increased computing power has made both permutation and bootstrap methods applicable to functional neuroimaging. We evaluate these approaches on t images using simulations and a collection of real datasets. We find that Bonferroni-related tests offer little improvement over Bonferroni, while the permutation method offers substantial improvement over the random field method for low smoothness and low degrees of freedom. We also show the limitations of trying to find an equivalent number of independent tests for an image of correlated test statistics.

Nonstationary cluster-size inference with random field and permutation methods

Because of their increased sensitivity to spatially extended signals, cluster-size tests are widely used to detect changes and activations in brain images. However, when images are nonstationary, the cluster-size distribution varies depending on local smoothness. Clusters tend to be large in smooth regions, resulting in increased false positives, while in rough regions, clusters tend to be small, resulting in decreased sensitivity. Worsley et al. proposed a random field theory (RFT) method that adjusts cluster sizes according to local roughness of images [Worsley, K.J., 2002. Nonstationary FWHM and its effect on statistical inference of fMRI data. Presented at the 8th International Conference on Functional Mapping of the Human Brain, June 2-6, 2002, Sendai, Japan. Available on CD-ROM in NeuroImage 16 (2) 779-780; Hum. Brain Mapp. 8 (1999) 98]. In this paper, we implement this method in a permutation test framework, which requires very few assumptions, is known to be exact [J. Cereb. Blood Flow Metab. 16 (1996) 7] and is robust [NeuroImage 20 (2003) 2343]. We compared our method to stationary permutation, stationary RFT, and nonstationary RFT methods. Using simulated data, we found that our permutation test performs well under any setting examined, whereas the nonstationary RFT test performs well only for smooth images under high df. We also found that the stationary RFT test becomes anticonservative under nonstationarity, while both nonstationary RFT and permutation tests remain valid under nonstationarity. On a real PET data set we found that, though the nonstationary tests have reduced sensitivity due to smoothness estimation variability, these tests have better sensitivity for clusters in rough regions compared to stationary cluster-size tests. We include a detailed and consolidated description of Worsley nonstationary RFT cluster-size test.

Validating cluster size inference: random field and permutation methods

Cluster size tests used in analyses of brain images can have more sensitivity compared to intensity based tests. The random field (RF) theory has been widely used in implementation of such tests, however the behavior of such tests is not well understood, especially when the RF assumptions are in doubt. In this paper, we carried out a simulation study of cluster size tests under varying smoothness, thresholds, and degrees of freedom, comparing RF performance to that of the permutation test, which is known to be exact. For Gaussian images, we find that the RF methods are generally conservative, especially for low smoothness and low threshold. For t images, the RF tests are found to be conservative at lower thresholds and do not perform well unless the threshold is high and images are sufficiently smooth. The permutation test performs well for any settings though the discreteness in cluster size must be accounted for. We make specific recommendations on when permutation tests are to be preferred to RF tests.

Oncologic Efficacy of CT-Guided Percutaneous Radiofrequency Ablation of Renal Cell Carcinomas

OBJECTIVE A single institutions experience with CT-guided percutaneous radiofrequency ablation of biopsy-proven renal cell carcinomas (RCCs) was studied to determine the disease-free survival and complication rate. MATERIALS AND METHODS Data from 125 RCCs in 104 patients treated with curative intent was reviewed. Radiofrequency ablation treatments were performed using conscious sedation and local anesthesia. Patients were followed with contrast-enhanced CT or MRI. Tumor control was defined as the absence of contrast enhancement in the tumor on CT or MRI. RESULTS Tumor size ranged from 0.6 to 8.8 cm (mean, 2.7 cm; SD, 1.5 cm). Of the 125 treated tumors, 116 (93%) were completely ablated (109 in a single ablation session, seven after a second ablation session) with a mean follow-up interval of 13.8 months. All 95 RCCs smaller than 3.7 cm were completely ablated, and 21 (70%) of 30 larger tumors were completely ablated, with nine showing evidence of residual viable tumor on follow-up scans. Tumor size smaller than 3.7 cm was significantly associated with achieving complete tumor eradication (p < 0.001). With each 1-cm increase in tumor diameter over 3.6 cm, the likelihood of tumor-free survival decreased by a factor of 2.19 (p < 0.001). There were 8 (8%) complications, none of which resulted in long-term morbidity. CONCLUSION CT-guided percutaneous radiofrequency ablation is a safe method to treat small RCCs. This study indicates that radiofrequency ablation can reliably eradicate RCCs smaller than 3.7 cm. Treatment of larger RCCs will result in an increased risk of residual RCC.

Combining voxel intensity and cluster extent with permutation test framework.

Voxel intensity-based tests provide good sensitivity for high intensity signals, whereas cluster extent-based tests are sensitive to spatially extended signals. To benefit from the strength of both, we consider combining intensity and extent information. We generalize previous work by proposing the use of weighted combining functions. Using a combining framework with permutation tests, we consider a variety of ways of combining voxel intensity and cluster extent information without knowing their distribution. Further, we propose meta-combining, a combining function of combining functions, which integrates strengths of multiple combining functions into one single statistic. Using real data, we demonstrate that combined tests can be more sensitive than voxel or cluster size test alone. Though not necessarily sensitive than individual combining functions, the meta-combining function is sensitive to all types of signals, thus can be uses as a single test summarizing all the combining functions.

Psychosocial Characteristics of Individuals With Non–Stage IV Melanoma

PURPOSE Melanoma is the fastest growing solid tumor among men and women and accounts for 79% of skin cancer-related deaths. Research has identified that distress is frequently associated with a diagnosis of cancer and may slow treatment-seeking and recovery, increasing morbidity and even mortality through faster disease course. Given that the 5-year survival rates for individuals with melanoma are determined primarily by the depth and extent of spread, distress that interferes with seeking treatment has the potential to be life-threatening. PATIENTS AND METHODS The current study was designed to identify levels of distress present in individuals seeking treatment at a large, Midwestern, multidisciplinary melanoma clinic. It also focused on determining the quality of life, level of anxiety, and coping strategies used by individuals with melanoma before treatment. Given that the course of treatment and outcome for patients with stage IV disease is vastly different from that of patients with stages I to III disease, they were excluded from the study. RESULTS Results indicated that most individuals who are presenting to a melanoma clinic do not report a clinically significant level of distress. However, there is some variability in this, with 29% of patients reporting moderate to high levels of distress. Moreover, analyses suggest that distressed individuals are more likely to use maladaptive coping strategies, such as escape-avoidance coping, and to have poorer quality of life. CONCLUSION Although most individuals do not present with significant levels of distress, a significant minority are distressed and rely more heavily on coping strategies that do not benefit them. Such individuals would likely benefit most from psychological intervention.

Whole mounted radical prostatectomy specimens do not increase detection of adverse pathological features.

PURPOSE The optimal method to process radical prostatectomy specimens to maximize the detection of adverse pathological features is unclear and accurate staging is critical. We compare the ability of whole mounted sections to detect these features compared to partially submitted radical prostatectomy specimens. MATERIALS AND METHODS A total of 93 consecutive radical prostatectomy specimens were processed as whole mounts. Tissue sections were analyzed and the pathological outcomes measured included Gleason score, surgical margin status, and presence or absence of extraprostatic tumor extension and/or seminal vesicle invasion. The pathological outcomes of the preceding cohort were compared to those of a similar cohort consisting of 554 men whose radical prostatectomy specimens were processed as partially submitted glands. RESULTS A multivariate logistic regression analysis was performed to determine the effect of the method of tissue processing on the pathological outcomes. When considered alone or adjusted for various preoperative patient characteristics (prostate specific antigen, biopsy Gleason score and clinical stage), there were no significant differences in the ability of whole mounted specimens to detect the various outcomes compared to partially submitted specimens (all p >0.4). CONCLUSIONS Whole mounted sampling of the radical prostatectomy specimen does not improve detection of adverse pathological features.

The regulation of prostate cancer cell adhesion to human bone marrow endothelial cell monolayers by androgen dihydrotestosterone and cytokines

A previous study from our laboratory suggested that prostate cancer metastasis to bone may be mediated, in part, by preferential adhesion to human bone marrow endothelial (HBME) cells. Tumor cell adhesion to endothelial cells may be modulated by the effect of cytokines on cell adhesion molecules (CAMs). Tumor necrosis factor-alpha (TNF-α) regulates VCAM expression on the endothelium and this effect is enhanced by dihydrotestosterone (DHT). Transforming growth factor-beta (TGF-β) stimulates the expression of α2β1integrin on PC-3 cells. The current study investigated the effects of the above cytokines and DHT (singularly and in various combinations) upon HBME and prostate cancer cell expression of VCAM, α2 integrin subunit, and β1 integrin subunit by flow cytometry. We also monitored the effects of the above treatments on PC-3 cell adhesion to HBME monolayers. The data demonstrate that none of the treatments significantly altered the expression of selected CAMs on HBME cell and neoplastic prostate cell lines. The treatment of HBME monolayers with various combinations of cytokines and DHT prior to performing adhesion assays with PC-3 demonstrates that treatments containing TGF-β reduced PC-3 cell adhesion to HBME monolayers by 32% or greater (P<0.05). The reduction in PC-3 cell adhesion to TGF-β-treated HBME monolayers was dose dependent. Interestingly, LNCaP cells but not PC-3 cells treated with TGF-β had a reduced ability to adhere to untreated HBME monolayers. These results suggest that TGF-β may reduce tumor cell adhesion to bone marrow microvascular endothelium, in vivo. The biological significance of this observation is discussed.

LACK OF ASSOCIATION OF PROSTATE CARCINOMA NUCLEAR GRADING WITH PROSTATE SPECIFIC ANTIGEN RECURRENCE AFTER RADICAL PROSTATECTOMY

PURPOSE Grading prostate cancer using the Gleason system relies only on architectural tumor growth, in contrast to other systems, such as the WHO system, which grade prostate carcinoma based on nuclear features as well as architectural patterns. The prognostic significance of nuclear grading remains controversial since most studies were performed before prostate specific antigen (PSA) screening became widely available. We evaluated the significance of nuclear grade for predicting PSA recurrence in a contemporary cohort of patients treated with radical prostatectomy for clinically localized prostate carcinoma. MATERIALS AND METHODS Nuclear grades 1 to 3 were determined in 141 consecutive radical prostatectomies in 1995. Predominant and worst nuclear grade was determined by a consensus of 3 pathologists. Statistical analysis compared nuclear grade with Gleason score using the chi-square test. The Cox proportional hazards analysis was performed to calculate the ability of nuclear grade, Gleason score and other variables to predict PSA recurrence. RESULTS We identified a significant association of Gleason score with worst nuclear grade (p = 0.007). All 6 cases with a Gleason score of 8 or greater had a worst nuclear grade of 3, in contrast to 36 of 60 (60%) with a score 6 or less, in which the worst nuclear grade was 3. Of the 141 patients 31 (21.9%) had PSA recurrence at a median followup of 3.7 years. The univariate Cox model revealed significant associations of PSA recurrence with Gleason score 8 or greater (hazards ratio 5.5, p = 0.005), extraprostatic extension (hazards ratio 3.4, p = 0.001), positive surgical margin (hazards ratio 2.6, p = 0.009), seminal vesicle involvement (hazards ratio 7.3, p <0.001), preoperative serum PSA (hazards ratio 1.03, p = 0.007), tumor stage (hazards ratio 3.6, p = 0.001) and maximal tumor dimension (hazards ratio 2.4, p <0.001). However, overall and worst nuclear grade did not predict PSA recurrence (p = 0.89 and 0.13, respectively). Nuclear grade did not fit any multivariate model tested, which otherwise included Gleason score, log(PSA), surgical margin status, extraprostatic extension, seminal vesicle status, tumor size and pathological stage. By varying sample fixation time we also showed that benign prostate tissue in the same section as prostate carcinoma had grade 2 or 3 nuclear changes, that is moderate to marked anaplasia. CONCLUSIONS High nuclear grade is associated with high Gleason score. However, prostate carcinoma with a Gleason score of 6 or less shows extreme variability. Nuclear grade determined by light microscopy failed to predict PSA recurrence in a contemporary series of men with clinically localized prostate cancer treated with radical prostatectomy. Nuclear morphology is subject to tissue fixation and processing artifact. Any nuclear morphometric study must consider this artifact.