Machine Learning for the Prediction of Biochemical Recurrence in Patients Treated with Radical Prostatectomy

PURPOSE/OBJECTIVE(S)

Biochemical recurrence (BCR) occurs in about 40% of patients with prostate cancer following radical prostatectomy (RP). Our goal was to develop a machine learning model for the prediction of BCR five-years after RP, to improve patient prognostication.

MATERIALS/METHODS

Patients treated with RP at a tertiary care medical center between 1990 and 2017 were included. A gradient boosted decision trees-based machine learning model modified to handle survival data was trained on 80% of the dataset. The model's performance was evaluated on the remaining 20%. Input variables were age at surgery, prostate specific antigen (PSA) at diagnosis (in ng/mL), pathologic Gleason grade group (GG), pathologic T stage (organ confined disease vs. extracapsular extension (ECE) vs. seminal vesicle invasion (SVI)), lymph node involvement, and surgical margin status. Model performance was assessed using time-dependent area under curve of the receiver operator curve (AUC).

RESULTS

The full dataset included 11,139 patients, of whom 1,153 (10%) developed BCR. Median age at surgery was 59 and PSA at diagnosis was 5.4 ng/mL. Only 1,080 (9.7%) patients had GG 3, and 707 (6.3%) GG 4 and 5. 1,366 (12%) patients had positive surgical margins and 134 (1.2%) had lymph node involvement. Most patients had organ confined disease with EPE and SVI diagnosed in 2,759 (25%) and 392 (3.5%) patients, respectively. Median follow-up was 5 years and median time to BCR was 4 years. When validated on the hold-out set of 2,228 patients, the model shows a time-dependent AUC of 0.82 (95% CI 0.78 - 0.86) for BCR at t = 5 years.

CONCLUSION

Our machine learning model can be used to estimate risk of BCR following RP and shows exceptional performance, with implications on patient prognostication and follow-up. We are currently working on validating its performance on an external dataset.

Multi-Institutional Analysis of Metastasis Directed Therapy with or without Androgen Deprivation Therapy in Oligometastatic Castration Sensitive Prostate Cancer

PURPOSE/OBJECTIVE(S)

Several prospective trials in oligometastatic castration sensitive prostate cancer (omCSPC) have shown metastasis-directed therapy (MDT) can delay time to progression and initiation of androgen deprivation therapy (ADT) compared to observation. However, the optimal integration of ADT with MDT remains unclear. Here we report a multi-national, multi-institutional retrospective cohort of omCSPC treated with MDT to characterize the long-term outcomes of patients treated with MDT alone or in combination with ADT.

MATERIALS/METHODS

Patients with a controlled primary site and omCSPC (defined as ≤ 5 lesions on conventional imaging) treated with MDT with or without concurrent ADT and with at least 36 months follow-up were retrospectively screened across 13 institutions. The primary endpoints included biochemical progression free survival (bPFS) and radiographic progression free survival (rPFS) calculated using Kaplan-Meier method and stratified by treatment group (MDT alone vs MDT + ADT). Multivariable Cox regression was performed adjusted for variables found to be prognostic on univariate analysis.

RESULTS

Among 414 patients screened, a total of 263 patients treated between 2003 and 2018 met inclusion criteria and included. Of these, 105 received MDT alone and 158 received MDT+ADT, with median follow-up of 49.5 and 54.5 months, respectively. The majority were metachronous (90%) and had bone lesions (60%). Median ADT duration was 21.3 months (IQR 12.0- 31.9). Patients who received ADT vs. no ADT had poorer prognostic features including 23% vs. 1% synchronous (p<0.001), and 55% vs 40% Gleason 8-10 (p = 0.012). ADT use was associated with a better 5-year bPFS 24% vs 11% (p<0.0001) and rPFS 41% vs 29% (p<0.001). On multivariable Cox regression adjusting for post-MDT PSA nadir and salvage therapy, ADT use maintained significance for both bPFS (HR 0.51 (0.36, 0.71), p<0.001) and rPFS (HR 0.67, 95% CI 0.46-0.96, p = 0.03).

CONCLUSION

Long-term outcomes with MDT alone suggest a small proportion of patients experience sustained disease control. The addition of ADT appears to improve rPFS, however prospective studies are needed in order to allow for personalization of care in patients with omCSPC.

Machine Learning for the Prediction of Adverse Pathological Outcomes in Patients Treated with Radical Prostatectomy

PURPOSE/OBJECTIVE(S)

Extracapsular extension (ECE) and seminal vesicle invasion (SVI) are associated with negative oncologic outcomes in patients with prostate cancer. We have developed and validated a machine learning model to more accurately identify patients at risk of these adverse surgical outcomes prior to radical prostatectomy (RP).

MATERIALS/METHODS

This study included a cohort of patients diagnosed with prostate cancer and treated with RP and lymph node dissection at a tertiary care medical center from 2010 to 2020. An ensemble model using a base gradient-boosted trees-based machine learning model and isotonic calibrators was trained on 80% of the cohort, with 20% held out for validation. The model uses age at surgery, prostate specific antigen level (PSA) at diagnosis, biopsy Gleason grade group, numbers of positive and negative cores on biopsy, and clinical T-stage (cT) as input variables. Model performance was assessed on the hold-out set using the area under the receiver operating curve (AUC).

RESULTS

The full dataset included 18,729 eligible patients. Median PSA at diagnosis was 7.3 ng/mL. Most patients had clinically organ confined disease (cT1 - cT2) with only 136 (0.7%) having cT3. The most common biopsy Gleason grade group was 2 (7,118 or 38% of patients), with Gleason grade 4 in 1,796 (9.6%), and 5 in 1,064 (5.7%) patients. After RP, 11,931 (64%) of patients had organ confined disease, 4,298 (23%) had ECE, and 2,500 (13%) had SVI. When validated on the hold-out set (n = 3,746), the model had AUCs of 0.79 (95%-CI 0.77 - 0.80), 0.67 (0.65 - 0.69), and 0.83 (0.81 - 0.85) for the prediction of organ confined disease, ECE, and SVI, respectively.

CONCLUSION

In conclusion, we have developed a machine learning model that predicts individual patient risk of pathologic T-stage. The model can be used to provide more accurate risk assessments and improve surgical treatment planning. We are currently working on externally validating our results on patients from different institutions.

Survival following early hormone therapy for men with rapid PSA doubling time within 2 years following radical prostatectomy

5065

Background: Early hormonal therapy has been used in the salvage setting for men with biochemical recurrence following radical prostatectomy (RP), but no studies to date have been able to evaluate whether such treatment prolongs survival. We examined the impact of salvage hormonal therapy on overall survival (OS) in a cohort with long-term follow-up, and attempted to identify the subgroup most likely to benefit. Methods: Retrospective analysis of a cohort of 488 men undergoing RP at Johns Hopkins Hospital from 1982–2004, who experienced biochemical recurrence and received no salvage therapy (n = 386) or salvage hormonal therapy (n = 102); no one received adjuvant therapy. Survival was defined from biochemical recurrence to death from all causes, and analyzed with proportional hazards models with time-dependent covariates. Results: With median follow-up of 6 years after recurrence and 9 years after RP, there were 143 deaths (29%), including 105 from prostate cancer. After adjusting for PSA doubling time (PSADT), RP Gleason score, and year of surgery, hormonal therapy did not significantly improve OS for all men, compared to no salvage therapy: hazard ratio (HR) = 0.72 (95% confidence interval (CI): 0.45–1.17), p = 0.187. However, when restricted to men with early recurrence, i.e. within 2 years of RP, and with a rapid PSADT<6 months, hormonal therapy was associated with a large, significant improvement in OS: HR = 0.25 (95% CI: 0.08–0.71), p = 0.0095. This subgroup comprised 22% of the cohort. In contrast, there was no benefit of salvage hormonal therapy in men with early recurrence and PSADT>6 months: HR = 1.96 (95% CI: 0.89–4.31), p = 0.093, nor those who recurred more than 2 years after RP, regardless of PSADT. Conclusions: This study suggests that early salvage hormonal therapy may significantly and substantially prolong overall survival in the subgroup of men who experience an early biochemical recurrence with a rapid PSADT. These results are consistent with early recurrences being indicative of metastatic disease, while later recurrences are more likely to represent local recurrence. If validated, these results may provide useful stratification criteria for clinical trials.

No significant financial relationships to disclose.

Multi-drug resistance and breast cancer: A meta-analysis of MDR1 and its functional significance

11010

Background: The clinical significance of MDR1 as a mechanism of multidrug resistance in breast cancer has not been established. We conducted a meta-analysis examining MDR1/gp170 in breast cancer, significantly updating a meta-analysis we performed in 1997. Methods: Published papers on MDR1/gp170 and breast cancer were identified by searching literature databases and bibliographies of published papers. Pooled relative risks (RR) for association between MDR1/gp170 expression and clinical outcomes were estimated with the Mantel-Haenszel method. Results: 84 studies were included in analysis: 63 measured MDR1/gp170 expression, 21 measured response to chemotherapy, 11 used sestamibi imaging, and 11 evaluated multidrug resistance reversing agents. MDR1 expression was detected in 45.9% of breast tumors prior to treatment and 66.8% of tumors after treatment (p<0.0001). There were no significant differences in gp170 expression by antibody type (p=0.567). In 11 studies, MDR1/gp170 expression after chemotherapy was associated with a significant risk of non-response, RR=1.79 (95% CI=1.35, 2.35), p<0.0001. This effect was lower in 13 studies with MDR1/gp170 assessed prior to chemotherapy, RR=1.28 (1.06, 1.55), p=0.009. Only 7 of 16 studies showed significant associations with decreased progression-free or overall survival. Of 11 sestamibi imaging studies, 10 showed a significant association between shorter retention and higher levels of MDR1/gp170, suggesting active protein pumping of contrast agent. None of 11 studies of multidrug reversing agents found >25% of refractory patients improving to partial clinical response, but most studies did not confine treatment to patients with MDR1/gp170 expression. Conclusions: This meta-analysis includes nearly 3 times as many studies as our 1997 analysis. MDR1 is expressed in approximately half of breast cancer patients, and is induced by chemotherapy. MDR1 after chemotherapy is associated with RR=1.8 for failure to achieve a clinical response, and RR=1.3 without induction by chemotherapy. Sestamibi gives the strongest support for active efflux of agent by gp170, but studies with MDR reversing agents have shown little success, which may partly reflect poor study design.

[Table: see text]

Breast cancer: dietary and environmental oestrogens

Breast cancer is among the most common of the cancers that occur in women living in western societies. It is a much-feared disease and the risks are confusing and often badly reported. Oestrogen levels are a known risk factor. But is it possible to lower the risk? And where are all the oestrogens coming from?

Predictive value of cathepsin-D for cervical lymph node metastasis in head and neck squamous cell carcinoma

BACKGROUND

Prognosis of head and neck squamous cell carcinoma (HNSCC) is strongly associated with cervical lymph node metastasis. Cathepsin-D is a lysosomal protease expressed in all cells. Its role in extracellular matrix degradation is postulated to promote tumor invasion and metastasis. Increased cathepsin-D has been demonstrated in cervical lymph node metastasis in HNSCC.

METHODS

Formalin fixed tumor biopsy samples from 34 patients with HNSCC of the oral cavity, oropharynx, or hypopharynx were analyzed for the presence of cathepsin-D by immunohistochemistry (1:8000, Calbiochem, Cambridge, MA). Tumors were considered positive if >50% of cells showed strong cytoplasmic staining.

RESULTS

All patients had T1 or T2 lesions ranging in size from 1-4 cm and 19 (56%) had cervical metastasis. Eight (24%) were well differentiated and 26 (76%) were moderately or poorly differentiated. Thirteen tumors (38%) had high cathepsin-D expression that was strongly associated with cervical lymph node metastasis (p = 0.008). When adjusted for tumor stage and grade, cathepsin-D positivity was nearly twice as likely to be associated with node metastasis (p = 0.011).

CONCLUSIONS

We demonstrated cathepsin-D expression in biopsies from a subset of patients with HNSCC and a strong association between this protease and cervical lymph node metastases. Cathepsin-D is a potential independent predictor of cervical lymph node metastasis in HNSCC and merits additional study.

Prognostic value of p53, glutathione S-transferase pi, and thymidylate synthase for neoadjuvant cisplatin-based chemotherapy in head and neck cancer

Neoadjuvant cisplatin-based chemotherapy has been widely used in the last decade for organ preservation or unresectable disease in advanced stage head and neck cancer. We examined the expression of a series of tumor markers that have been associated with chemotherapy resistance in pretreatment biopsies from 68 patients who received cisplatin-based neoadjuvant chemotherapy at either of two institutions. Patients received either cisplatin/5-fluorouracil (n = 49) or cisplatin/paclitaxel (n = 19). Expression of p53, glutathione S-transferase pi (GSTpi), thymidylate synthase (TS), c-erbB2, and multidrug resistance-associated protein was examined by immunohistochemistry. Expression of glutathione synthetase mRNA was measured by in situ hybridization. The overall response rate for cisplatin-based neoadjuvant treatment was 79%. The expression of several of the tumor markers was associated with resistance to neoadjuvant treatment, but none reached statistical significance. Overall survival (OS) was strongly correlated with the absence of p53 expression. The OS at 3 years was 81% in the p53-negative group, whereas it was 30% in the p53-positive group for patients treated with neoadjuvant chemotherapy (P < 0.0001). Expression of GST pi and TS was also significantly correlated with decreased OS after neoadjuvant treatment. At 3 years, the OS rate was 82% in the low GSTpi score group, compared to 46% in the high GSTpi score group (P = 0.0018). In the TS-negative group, the 3-year OS rate was 71% compared with 40% in the TS-positive group (P = 0.0071). We conclude that p53, GSTpi, and TS may be clinically important predictors of survival in patients receiving neoadjuvant chemotherapy for head and neck cancer.

Assessing the clinical impact of prognostic factors: when is "statistically significant" clinically useful?

Very few tumor markers have been recommended for routine clinical care of patients with breast cancer. A framework to determine the clinical utility of tumor markers is required. In a previous publication, a "Tumor Marker Utility Grading System" (TMUGS) was proposed. TMUGS included a semi-quantitative grading scale (0-3+) which can be used to assign a score to a given tumor marker for a given outcome. Only those markers that are felt to be sufficiently strong to influence a therapeutic decision that results in improved clinical outcome for the patient are recommended. The studies from which data are used to assign a TMUGS grade can be placed into one of five Levels of Evidence (LOE). An extension of TMUGS ("TMUGS-Plus") is now proposed in which the relative strength of a prognostic or predictive factor can be estimated and expressed in terms of a risk ratio (RR) for prognostic factors or benefit ratio (BR) for predictive factors. Three categories of prognostic factors and three categories of predictive factors are proposed (strong, moderate, and weak). It is recommended that only LOE type I studies (prospective, highly powered studies of the tumor marker, or meta-analysis of LOE II or III datasets), be used to estimate the RR or BR of a given factor. Finally, a matrix, based on assumptions of acceptable absolute benefits relative to risks, is proposed in which any given tumor marker can be assessed for its clinical utility. TMUGS-Plus should aid in the assessment of published data regarding clinical utility of tumor markers. Perhaps more important, clinical investigators can use TMUGS-Plus to design tumor marker studies that will fulfill criteria for clinical utility, resulting in more rapid acceptance of tumor markers for routine clinical use.

Autoantibodies to the nuclear phosphoprotein nucleophosmin in breast cancer patients

Nucleophosmin (NPM) is an estrogen-regulated nucleolar phosphoprotein; a substrate for phosphorylation by p34cdc2 kinase, protein kinase C, and casein kinase II; and a repressor of the transcriptional regulating activities of the YY1 and IFN regulatory factor-1 transcription factors. We have completed a pilot study to determine whether autoantibodies to NPM are present in breast cancer patients and explored the ability of these autoantibodies to predict recurrence in breast cancer patients. One hundred breast cancer patients were studied: 50 who recurred, and 50 matched for age and length of follow-up but who did not recur. Patients' sera were collected at the times of diagnosis (T1), six months before recurrence (T2), and at recurrence (T3). Recurrent and nonrecurrent patients did not differ in autoantibody levels at the times of diagnosis or recurrence. However, antiNPM autoantibody levels increase significantly between diagnosis and six months before recurrence in recurrent patients, whereas no change occurs over the comparable time period in nonrecurrent patients (repeated measures ANOVA; P = 0.041). At recurrence, the levels return to those seen at diagnosis. The greater the change in levels between T1 and T2, the greater the risk of recurrence within the next 6 months (conditional logistic regression: increase in risk for highest versus lowest tertile of change from T1 to T2; odds ratio, 3.25; 95% confidence interval, 1.04-10.18; P = 0.043). Consistent with the estrogenic/antiestrogenic regulation of the antigen in breast cancer cells, the levels of antiNPM autoantibodies are decreased 6 months before recurrence in patients treated with the antiestrogen tamoxifen (P = 0.012). The association between antiNPM levels and recurrence remained after adjustment for confounding factors. Further study of antiNPM autoantibody levels as a new and simple, intermediate serum biomarker for predicting both the timing of recurrence and monitoring response to endocrine manipulations in breast cancer patients is warranted.

Stromal IGF-II messenger RNA in breast cancer: relationship with progesterone receptor expressed by malignant epithelial cells

In breast cancer, insulin-like growth factor II (IGF-II) is stromal in origin and is considered an important regulator of tumour epithelium growth. The presence of progesterone receptor (PR) is expression of an intact oestrogen regulatory pathway of breast malignant epithelial cells and represents a parameter of cell differentiation in breast cancer. In this study we have examined the relationship between IGF-II mRNA expression and ER, PR content in 75 breast cancer. Formalin-fixed paraffin-embedded tissue sections were used to preserve histological details. IGF-II mRNA was evaluated by in situ hybridisation method and ER, PR by immunohistochemistry. IGF-II mRNA was scored semi-quantitatively: 2.6% breast tumour specimen expressed no IGF-II mRNA, 46.7% had low levels of expression (IGF-II-) and 50.7% had moderate or high IGF-II mRNA content (IGF-II+). IGF-II mRNA was found in the stroma fibroblasts surrounding malignant lesions and no signal was detected in malignant epithelial cells. In contrast, ER and PR were expressed only by neoplastic epithelial cells and no immunoreactivity was found in the stroma: 50/75 (66.6%) breast cancer specimens were positive for ER (ER+) and 35 (46.6%) for PR (PR+). Both, IGF-II mRNA and PR were directly correlated with the stromal proliferation (p < 0.05 and p < 0.001, respectively). No relationship was found between IGF-II RNA and ER. In contrast 24/35 (73.5%) PR breast cancer tissues were IGF-II+ (p < 0.01) and a strong correlation was found between epithelial PR immunostaining and stromal IGF-II mRNA content (p < 0.003). Our data indicate that in breast cancer IGF-II mRNA is generally expressed by stromal cells and ER and PR by epithelial cancer cells, and that IGF-II mRNA expression is strongly related with both percentage and staining intensity of PR+ epithelial cancer cells. These data support the hypothesis that IGF-II produced by the fibroblasts may exert a paracrin effect on malignant epithelium regulating its differentiation.

BRCA1 testing in families with hereditary breast-ovarian cancer. A prospective study of patient decision making and outcomes

OBJECTIVES

To identify predictors of utilization of breast-ovarian cancer susceptibility (BRCA1 gene) testing and to evaluate outcomes of participation in a testing program.

DESIGN

Prospective cohort study with baseline interview assessment of predictor variables (eg, sociodemographic factors, knowledge about hereditary cancer and genetic testing, perceptions of testing benefits, limitations, and risks). BRCA1 test results were offered after an education and counseling session in a research setting. Outcome variables (including depression, functional health status, and prophylactic surgery plans [follow-up only]) were assessed at baseline and 1-month follow-up interviews.

PARTICIPANTS

Adult male and female members (n=279) of families with BRCA1-linked hereditary breast-ovarian cancer (HBOC).

RESULTS

Of subjects who completed a baseline interview (n=192), 60% requested BRCA1 test results (43% of all study subjects requested results). Requests for results were more frequent for persons with health insurance (odds ration [OR], 3.74; 95% confidence interval [CI], 2.06-6.80); more first-degree relatives affected with breast cancer (OR, 1.59; 95% CI, 1.16-2.16); more knowledge about BRCA1 testing (OR, 1.85; 95% CI, 1.36-2.50); and indicating that test benefits are important (OR, 1.45; 95% CI, 1.13-1.86). At follow-up, noncarriers of BRCA1 mutations showed statistically significant reductions in depressive symptoms and functional impairment compared with carriers and nontested individuals. Individuals identified as mutation carriers did not exhibit increases in depression and functional impairment. Among unaffected women with no prior prophylactic surgery, 17% of carriers (2/12) intended to have mastectomies and 33% (4/12) to have oophorectomies.

CONCLUSIONS

Only a subset of HBOC family members are likely to request BRCA1 testing when available. Rates of test use may be higher in persons of a higher socioeconomic status and those with more relatives affected with breast cancer. For some high-risk individuals who receive test results in a research setting that includes counseling, there may be psychological benefits. More research is needed to assess the generalizability of these results and evaluate the long-term consequences of BRCA1 testing.

Familial uveal melanoma. Clinical observations on 56 patients

OBJECTIVE

To study the clinical profile and kindreds of patients with familial uveal melanoma (FUM).

DESIGN

Retrospective case series.

SETTING

Tertiary referral center.

PATIENTS

Medical charts of 4500 patients with uveal melanoma were reviewed for family history of uveal melanoma. The clinical profile of these patients and their kindreds were studied to determine the incidence of FUM and pattern of inheritance. The association of FUM to cutaneous melanoma, familial atypical mole and melanoma syndrome, and other nonmelanocytic cancers was analyzed using statistical methods.

RESULTS

Of 4500 patients with uveal melanoma, 56 patients in 27 families (0.6%) had a family history of uveal melanoma. The uveal melanoma in all 56 familial patients was unilateral. In 17 cases (63%), the second affected relative was a first-degree relative. In the remainder, the second affected relative was a second- (22%) and third-degree (15%) relative. In 25 families (93%) only two members were affected, and in two families (7%) three members had uveal melanoma. Patients with FUM were four times as likely to have a second primary malignant neoplasm than were people in the general population. However, no evidence was seem that unaffected kindreds of patients with FUM were at higher risk of having a second primary malignant neoplasm.

CONCLUSIONS

Familial involvement in uveal melanoma is rare. Familial uveal melanoma most often (63%) affects first-degree relatives, rarely affects more than two persons in a family, and may be associated with a generalized inherited predisposition to cancer. Further genetic studies are necessary to fully characterize FUM syndrome.

Phase II study of oral etoposide for patients with advanced breast cancer

BACKGROUND

The objective of this study was to define the activity and toxicity of etoposide for patients with previously treated metastatic breast cancer.

METHODS

Thirty patients with measurable metastatic breast cancer who progressed after prior chemotherapy, hormonal therapy, or both, either as adjuvant therapy or for metastatic disease, were enrolled. There were 26 patients evaluable for response and 30 evaluable for toxicity. Treatment consisted of oral etoposide at a dose of 50 mg/m2/day administered for 21 consecutive days. Doses were modified depending on toxicity. Patients were evaluable for response after at least one cycle of therapy.

RESULTS

One complete response and four partial responses were observed. The overall objective response rate was 19% (5/26). Six patients had stable disease. Toxicity consisted mainly of Grade 4 neutropenia for 24% and Grade 4 thrombocytopenia for 13% of patients. There was one death due to neutropenic sepsis.

CONCLUSION

Oral etoposide is an effective, easily administered outpatient regimen with minimal toxicities. Etoposide is effective for patients with pretreated metastatic breast cancer.

Overexpression of the p53 gene in primary and metastatic head and neck carcinomas

The p53 gene has been correlated with disease progression in a number of human malignancies, and p53 abnormalities are found in a high percentage of head and neck squamous cell carcinomas. The objectives of this study were 1. to correlate p53 expression with disease progression in squamous cell carcinoma of the head and neck (SCCHN), and 2. to determine whether there are site-specific differences in p53 expression. Primary lesions and/or lymph node metastases from 147 patients with invasive SCCHN were immunostained for p53 overexpression. Expression of p53 was similar (42% versus 43%) in primary lesions and lymph node metastases. Expression also did not vary significantly by site in the head and neck. In conclusion, increased p53 expression did not correlate with disease progression in our series of patients with invasive SCCHN. The finding of a lack of increased expression with disease spread to lymph nodes supports the belief that p53 alterations occur early in head and neck carcinogenesis.

The optimal temporal window for CT of the liver using a time-density analysis: implications for helical (spiral) CT

OBJECTIVE

Scanning protocols for conventional CT of the liver have been proposed. Current availability of helical CT with a four- to sixfold decrease in scan time requires significant adjustments in these protocols. The present study assesses the implications of time-density curves on the performance of helical liver CT.

MATERIALS AND METHODS

Twenty patients without liver lesions were studied for time-density analysis of the aorta, inferior vena cava (ICV), portal vein, and liver. Scans were performed at the level of the portal vein at baseline and every 15 s for 3 min following uniphasic administration of 150 ml (300 mg I/ml) nonionic contrast agent. Regions of interest were used to measure three areas in each anatomic structure over time. Median and mean peak enhancement times were calculated for all 20 patients. Cubic spline interpretation was employed to determine the point of equilibrium.

RESULTS

Results demonstrated the following average maximum enhancement values and times for peak enhancement: aorta: 227 HU (75 s); liver: 123 HU (105 s); portal vein: 187 HU (90 s); IVC: 142 HU (90 s). Hepatic enhancement achieved 67 HU over baseline. Peak portal enhancement occurred 15 s prior to liver enhancement (p = 0.001). Aortic and hepatic curves became parallel (onset of equilibrium) at a median time of 120 s.

CONCLUSION

Helical scanning requires a longer delay (70-80 s) than used for conventional CT. Upon application of these principles, scan initiation occurs higher on the liver enhancement curve, improving liver enhancement without impinging on equilibrium.

p53 is overexpressed in fifty percent of pre-invasive lesions of head and neck epithelium

To identify the stage in head and neck carcinogenesis at which p53 abnormalities become important, we evaluated 512 squamous epithelial tissue samples (201 pre-invasive and 209 invasive lesions, as well as 102 normal epithelia) in specimens from two institutions. Of 311 patients evaluated, 128 did not have an invasive carcinoma. The frequency of p53 overexpression in the pre-invasive lesions was not influenced by an adjacent invasive tumor. There was an increasing frequency of p53 overexpression in specimens from histologically normal epithelium (5%, with only scattered basal layers positive) to mild (28%) and moderate dysplasia (47%). p53 overexpression was also significantly more common in lesions with severe dysplasia (54%) and carcinoma in situ (CIS; 50%) than in those with mild dysplasia. To further evaluate the timing of p53 alterations in the development of head and neck cancer, paired epithelial samples (one pre-invasive and one invasive) from the same patient were evaluated. A discordance in the p53 staining pattern was seen in approximately one-third of the cases. When a discordance was present, the frequency with which the invasive lesion was positive but the non-invasive negative decreased from 97% (invasive versus normal epithelium) to 50% (invasive versus severe dysplasia or CIS). When p53 expression was evaluated by site, buccal lesions appeared to overexpress p53 more frequently than tongue lesions. In conclusion, p53 protein overexpression is an early event in head and neck carcinogenesis and may represent a biomarker for patients with pre-invasive lesions.

Effects of topical tretinoin on dysplastic nevi

PURPOSE

As potential precursors of melanoma and markers of increased melanoma risk, dysplastic nevi are suitable targets of strategies for melanoma chemoprevention. We report the results of a pilot study of topical retinoic acid in patients with dysplastic nevi.

PATIENTS AND METHODS

Five male patients with dysplastic nevi applied tretinoin to half of the back for 6 months. Baseline photographs of dysplastic nevi were compared with posttreatment photographs and assessed for morphologic change. At study completion, each subject had four nevi excised from the treated side and four from the untreated side of the back. Biopsies were histologically evaluated for the presence of dysplasia.

RESULTS

All patients developed signs of irritation as a result of treatment. One patient was not compliant with treatment due to skin irritation. The four compliant patients showed significant decreases in the clinical atypia of treated lesions, with concomitant fading and even disappearance of many treated nevi. Histologically, only four of 16 treated nevi met histologic criteria for dysplasia, in comparison to 13 of 16 untreated nevi.

CONCLUSION

These results suggest that there is concomitant clinical and histologic improvement in a significant percentage of dysplastic nevi treated with topical tretinoin. However, the utility of topical tretinoin for chemoprevention of melanoma is limited by difficulty of application and associated inflammation. While new strategies in chemoprevention of melanoma are explored, sun protection and assiduous avoidance of sunburn must remain the mainstay of melanoma prevention.

Natural history of dysplastic nevi

BACKGROUND

Dysplastic nevi are markers of melanoma risk and potential precursors of melanoma. Few studies have addressed the natural history of dysplastic nevi.

OBJECTIVE

Our purpose was to describe the changes observed in nevi over time in a cohort of our patients with dysplastic nevi.

METHODS

We used a historical cohort design to study 153 patients with dysplastic nevi observed for a minimum of 5 years. Physical examination at completion of the study was compared with baseline overview and close-up photographs. Lesions excised during the study interval were assessed for history of change.

RESULTS

Fifty one percent of all evaluated nevi (297 of 593) showed clinical signs of change during an average follow-up of 89 months. New nevi were common in adulthood and continued to form in more than 20% of patients older than 50 years of age. Nevi were observed to become more clinically atypical, less clinically atypical, and disappear in all age groups. In this small cohort rates of nevus change were not correlated with personal or family history of melanoma, sex, or total number of nevi. Total nevus counts and rates of nevus change were correlated with age.

CONCLUSION

Dysplastic nevi remain clinically dynamic in adulthood. Our data suggest that the decrease in counts of dysplastic nevi associated with increasing age is only partly explained by the disappearance of nevi over time and probably reflects a tendency to larger numbers of nevi among more recent birth cohorts.

A cohort study of melanoma in patients with dysplastic nevi

A historical cohort of 153 patients with dysplastic nevi was studied for the development of melanoma. Each subject had a minimum follow-up of 5 years, with an average study follow-up of 94 months. Eleven new melanomas developed in 11 patients between 11 and 143 months (average 61 months) into the study period. Age-adjusted melanoma incidence in the cohort was 692/100,000 person years. Development of melanoma was strongly correlated with prior personal and/or family history of melanoma. Even among the 89 patients with no personal or family history of melanoma, the age-adjusted incidence of melanoma was an alarming 154/100,000 person years. Within this small, non-randomly ascertained cohort no association could be discerned between melanoma incidence and total number of banal or dysplastic nevi. These findings support the public health significance of the recognition of dysplastic nevi and confirm the importance of family history in evaluating melanoma risk associated with dysplastic nevi.

Impact of an HMO-based intervention to increase mammography utilization

A health maintenance organization (HMO)-based program designed to increase breast cancer screening was evaluated, focusing on changes in mammography utilization. The program consisted of a multistage intervention aimed at women members and primary care physicians of the HMO. This report examines the effect of the intervention on mammography utilization. The program was evaluated using a quasiexperimental design in which a random sample of women aged 50-74 from the HMO (intervention) was compared to a similarly aged geographic control group selected through random digit dialing. From 1988 to 1990, 450 intervention women and 450 control women were sampled (without replacement) each year and surveyed about breast cancer screening practices and related knowledge. A clear increase in self-reported mammography utilization was associated with the intervention. The percentage of women who reported a mammogram in the 12 months prior to the survey increased from 41% in 1988 (baseline) to 68% in 1990 among HMO women, compared to a change from 39% to 49% among control women. Comparing postintervention rates of mammography in HMO versus control women yielded a rate ratio (RR) of 1.4. However, this effect was strongly modified by income and race. Women with annual incomes of $31,000 or more showed little (whites, RR = 1.2) or no (blacks, RR = 1.0) effect of the intervention. Among women with incomes less than $31,000, the effect among whites (RR = 1.9) was much stronger than among blacks (RR = 1.2).

The logistic modeling of sensitivity, specificity, and predictive value of a diagnostic test

A method is described for modeling the sensitivity, specificity, and positive and negative predictive values of a diagnostic test. To model sensitivity and specificity, the dependent variable (Y) is defined to be the dichotomous results of the screening test, and the presence or absence of disease, as defined by the "gold standard", is included as a binary explanatory variable (X1), along with variables used to define the subgroups of interest. The sensitivity of the screening test may then be estimated using logistic regression procedures. Modeled estimates of the specificity and predictive values of the screening test may be similarly derived. Using data from a population-based study of peripheral arterial disease, the authors demonstrated empirically that this method may be useful for obtaining smoothed estimates of sensitivity, specificity, and predictive values. As an extension of this method, an approach to the modeling of the relative sensitivity of two screening tests is described, using data from a study of screening procedures for colorectal disease as an example.

Psychological and behavioral implications of abnormal mammograms

OBJECTIVE

To evaluate women's psychological responses to abnormal mammograms and the effect on mammography adherence. To identify psychological responses and other factors that predict mammography adherence in women with normal or abnormal mammograms.

DESIGN

Survey study with prospective analysis of factors associated with mammography adherence.

SETTING

Health Maintenance Organization of Pennsylvania and New Jersey (HMO PA/NJ).

PATIENTS

Study patients, members of HMO PA/NJ who were 50 years of age or older, and who had had mammography done 3 months earlier, included women with normal mammograms (n = 121), women with low-suspicion mammograms (n = 119), and women with high-suspicion mammograms (n = 68), but not women with breast cancer.

MEASUREMENTS

Psychological responses 3 months after mammography and adherence to subsequent annual mammography were assessed.

MAIN RESULTS

Women with high-suspicion mammograms had substantial mammography-related anxiety (47%) and worries about breast cancer (41%). Such worries affected the moods (26%) and daily functioning (17%) of these women, despite diagnostic evaluation excluding malignancy. For each variable, a consistent trend (P greater than 0.05) was seen with degree of mammogram abnormality. Sixty-eight percent of women with normal results, 78% of women with low-suspicion results, and 74% of women with high-suspicion results obtained their subsequent annual mammograms (P greater than 0.05). The number of previous mammograms (odds ratio, 3.2; 95% CI, 1.6 to 6.2) and the effect of the previous results on concerns about breast cancer (odds ratio, 0.5; CI, 0.2 to 1.0) were independent predictors of adherence in logistic regression analyses (P less than 0.05).

CONCLUSIONS

A substantial proportion of women with suspicious mammograms have psychological difficulties, even after learning that they do not have cancer. Such sequelae do not appear to interfere with subsequent adherence.

Why do some women get regular mammograms?

One in nine American women will be diagnosed with breast cancer at some time in their lives. Although mammography is a proven technology for diagnosing early, curable breast cancer, most women do not obtain regular mammograms. The purpose of this study was to identify the characteristics of women who received more than one mammogram. Such information is needed to develop health education programs aimed at fostering adherence to routine mammographic screening. The data were obtained from a 1989 random telephone survey of 910 women 50-74 years of age. The survey is part of the evaluation strategy for a five-year breast screening study. Two-thirds of the respondents reported having had mammograms. Most of these women had one or two mammograms. We report two logistic regression models which describe women who had mammograms in the past year and those who had one versus two or more mammograms. The most important variable in the models was physician support. Other consistently significant variables included being willing to pay $75-$100 (although most women did not pay this amount), visiting the doctor at least annually when healthy, being a nonsmoker, and recognizing that women older than 50 are at greater risk for breast cancer. The results indicate that two-pronged educational strategies aimed at both women and their physicians are needed to increase the percentage of women who obtain regular mammograms.

Regression methods for estimating attributable risk in population-based case-control studies: a comparison of additive and multiplicative models

A regression method that utilizes an additive model is proposed for the estimation of attributable risk in case-control studies carried out in defined populations. In contrast to previous multivariate procedures for the estimation of attributable risk, which have utilized logistic regression techniques to adjust for confounding factors, the model assumes an additive relation between the covariates included in the regression equation. As an empirical example, additive and logistic models were fitted to matched case-control data from a population-based study of childhood astrocytoma brain tumors. Although both models fitted the data well, the additive model provided a more satisfactory estimate of the risk attributable to multiple exposures, in the absence of significant additive interaction. In contrast to the results from the logistic model, the adjusted estimates of the risk attributable to each factor included in the additive model summed to the overall estimate for all of the factors considered jointly. Thus, the additive approach provides a useful alternative to existing procedures for the multivariate estimation of attributable risk when the additive model is determined to be appropriate on the basis of goodness-of-fit.

Psychological side effects of breast cancer screening

Evaluated the impact of receiving abnormal mammogram results on women's anxiety and breast cancer worries and on their breast self-examination (BSE) frequency and intentions to obtain subsequent mammograms. A telephone survey was conducted with 308 women 50 years old and older approximately 3 months following a screening mammogram. Subjects included women with suspicious abnormal mammograms, nonsuspicious abnormal mammograms, and normal mammograms. Women with suspicious abnormal mammograms exhibited significantly elevated levels of mammography-related anxiety and breast cancer worries that interfered with their moods and functioning, despite the fact that diagnostic work-ups had ruled out breast cancer. Women with moderate levels of impairment in mood or functioning were more likely to practice monthly BSE than women with either high or low levels of impairment. Breast cancer worries, perceived susceptibility to breast cancer, and physician encouragement to get mammograms all exhibited independent positive relationships to mammogram intentions.

How valid are mammography self-reports?

We compared mammography reports in medical records to self-reports obtained during a 1989 telephone interview survey for a sample of 100 women members of a health maintenance organization (HMO) who indicated they had mammograms within the past year and 100 who said they had not had mammograms within the past year. Of the women reporting they had not had mammograms within the past year, none had mammogram reports in the HMO data center. Of the 100 women reporting they had mammograms within the past year, 94 had confirmatory radiology records.

Factors associated with repeat adherence to breast cancer screening

This study identified barriers and facilitators of repeat participation in mammography and breast physical examination among women ages 50 years and over. Telephone interviews were conducted with 910 women in this age group. Forty percent of respondents had never had a mammogram. Only 38% had had one in the past 12 months. Of women who had a prior mammogram, 43% had had only one. Only 60% of women had had a breast exam in the past 12 months. A physician recommendation was the single best predictor of adherence to mammography. However, only 60% of women reported that their physicians had ever recommended mammography. Several other barriers to mammography were revealed, including anxiety, embarrassment, and concerns about cost and radiation. Both a family history of breast cancer and heightened perceived vulnerability to breast cancer were associated positively with repeat mammography participation; anxiety about screening reduced the likelihood of this outcome. These findings suggest that physicians can play a powerful role in motivating women to participate in initial and subsequent breast cancer screening. Reassurance may reduce women's anxiety and embarrassment and increase utilization further.

Dietary fiber, vegetables, and colon cancer: critical review and meta-analyses of the epidemiologic evidence

Whether colon cancer risk can be modified by a diet rich in vegetables, grains, and fruit, and, if so, whether the protective factor is dietary fiber or other biologically active components correlated with a high-fiber diet are questions of active research interest. Because studies on diet are susceptible to bias from a number of sources, in this review we evaluated the adequacy of study methodology as well as study results to clarify how much protection, if any, is conferred by a high-fiber diet. The review consisted of an aggregate assessment of the strength of evidence from 37 observational epidemiologic studies as well as meta-analyses of data from 16 of the 23 case-control studies. Both types of analyses revealed that the majority of studies gave support for a protective effect associated with fiber-rich diets; an estimated combined odds ratio (OR) of 0.57 (95% confidence interval = 0.50, 0.64) was obtained when the highest and lowest quantiles of intake were compared. Risk estimates based on vegetable consumption (OR = 0.48) were only slightly more convincing than those based on an estimate of fiber intake (OR = 0.58), but the data do not permit discrimination between effects due to fiber and nonfiber effects due to vegetables.

Breast screening practices among primary physicians: reality and potential

Increased use of regular screening mammograms and clinical breast examinations (CBE) among women aged 40 years and more could have a dramatic impact on mortality from breast cancer, but patient and physician barriers to mammography impede its acceptance. We conducted a survey of 300 primary care physicians to assess their knowledge, attitudes, beliefs, and breast screening practices. Our results show that only 71 percent of the respondents ordered mammograms for all women aged 50 to 75 years, which is the recommendation by the National Cancer Institute and American Cancer Society. Approximately 46 percent of respondents performed CBE on all women patients aged 50 to 75 years. Inadequate patient insurance coverage, equivocal radiology reports, patient reluctance or worry, and patient embarrassment all appear to be barriers to physicians' utilization of breast screening.

Psychological side effects of breast cancer screening

Evaluated the impact of receiving abnormal mammogram results on women's anxiety and breast cancer worries and on their breast self-examination (BSE) frequency and intentions to obtain subsequent mammograms. A telephone survey was conducted with 308 women 50 years old and older approximately 3 months following a screening mammogram. Subjects included women with suspicious abnormal mammograms, nonsuspicious abnormal mammograms, and normal mammograms. Women with suspicious abnormal mammograms exhibited significantly elevated levels of mammography-related anxiety and breast cancer worries that interfered with their moods and functioning, despite the fact that diagnostic work-ups had ruled out breast cancer. Women with moderate levels of impairment in mood or functioning were more likely to practice monthly BSE than women with either high or low levels of impairment. Breast cancer worries, perceived susceptibility to breast cancer, and physician encouragement to get mammograms all exhibited independent positive relationships to mammogram intentions.