An attempt to independently verify the utility of the Van Nuys Prognostic Index for ductal carcinoma in situ

Authors


  • Presented in part as an oral presentation at the American Society for Therapeutic Radiology and Oncology (ASTRO) 47th Annual Meeting, Denver, Colorado, October 16–19, 2005. An attempt to independently verify the utility of the Van Nuys Prognostic Index (VNPI) for DCIS. Int J Radiat Oncol Biol Phys. 2005;63(suppl 1):S57-S58.

Abstract

BACKGROUND.

The Van Nuys Prognostic Index (VNPI) purports to predict the risk of ipsilateral breast tumor recurrence (IBTR) after excision of ductal carcinoma in situ (DCIS). It is a simple scoring scheme based on a retrospective evaluation of data from a single group of investigators. Various versions of VNPI have been proposed using clinical and pathologic features including tumor size, tumor grade, margin width, and patient age. Despite common use of VNPI in the clinical management of patients with DCIS, independent validation is lacking.

METHODS.

A total of 222 patients were retrospectively analyzed with mammographically detected DCIS who were treated with surgical excision alone. Wire-localized excisional biopsy was performed and surgical specimens were measured and inked to assist in margin assessment. Multiple sections of each specimen were evaluated for histopathologic subtype, histologic and nuclear grade, presence of necrosis, maximum dimension of the lesion, and margin width. Each patient was prospectively evaluated by a multidisciplinary management team and presented with adjuvant treatment options including whole breast radiotherapy and/or tamoxifen. All patients in this cohort declined radiotherapy. Thirty-one percent of patients received tamoxifen. Patients were followed clinically every 3 to 6 months, and mammographically every 6 to 12 months. IBTR was confirmed by biopsy. Wilcoxon regression analysis was used to evaluate risk groups according to 3 proposed VNPI classification schemes: VNPI Group 1 (margin, grade, and size), VNPI Group 2 (margin, grade, size, and patient age), and VNPI Group 3 (margin only).

RESULTS.

With a median follow-up of 4.6 years, the crude rate of IBTR was 8.6% for the entire cohort. Of the patients who developed an IBTR, 73.7% had a lesion with a maximum dimension of ≤15 mm, 47.4% had a margin ≥10 mm, and 36.8% had grade 1 histology. At 5 years, IBTR was statistically indistinguishable for the 3 VNPI models. The 5-year freedom from IBTR for low-risk, intermediate-risk, and high-risk groups according to VNPI Group 1 was 96%, 84%, and 100%, respectively (P = .20). Similarly, the 5-year freedom from IBTR for low-risk, intermediate-risk, and high-risk groups according to VNPI Group 2 was 95%, 83%, and 100%, respectively (P = .19). Taking into account margin status only (VNPI Group 3) the 5-year freedom from IBTR for low-risk, intermediate-risk, and high-risk groups was 92%, 91%, and 91%, respectively (P = .98). Tamoxifen use did not appear to affect the 5-year rate of IBTR (95% vs 94%; P = 1.0).

CONCLUSIONS.

The results of the current study suggest that VNPI or margin width alone is not a valid tool with which to assist in the stratification of patients after excision alone for their risk of IBTR at 5 years. Further follow-up may strengthen the predictive utility of the various VNPI classification schemes. Cancer 2007. © 2007 American Cancer Society.

Ductal carcinoma in situ (DCIS) is a heterogeneous neoplastic process confined to the ductal system. With the advent of mammography, the rate of detection of DCIS has risen precipitously. From 1973 to 1992, the incidence of DCIS increased a remarkable 557% compared with a 34% increase in the diagnosis of invasive breast cancer.1 Currently, DCIS represents approximately 25% of newly diagnosed breast cancer cases.2

A significant proportion of patients diagnosed with DCIS can be appropriately treated with breast conservation therapy (BCT). Many have attempted to identify risk factors for ipsilateral breast tumor recurrence (IBTR) after wide local tumor excision with or without radiotherapy for DCIS. Most studies have demonstrated an association between nuclear grade and/or necrosis and the risk of IBTR.3–5 Other studies have shown that the status of the microscopic surgical margin and patient age are significant predictive variables.6–10

Silverstein et al.11 have proposed the Van Nuys Prognostic Index (VNPI) as a means to stratify patients with DCIS into low-risk, intermediate-risk, and high-risk groups. Based on retrospective analysis of data from this single group of investigators, each patient is assigned a score from 1 to 3 based on the grade, size, and margin width (Table 1). Patients with a cumulative score of 3 to 4 are considered ‘low-risk’ and are recommended for excision alone, a score of 5 to 7 is considered to be ‘intermediate-risk,’ and suitable for excision with radiotherapy, and patients with a ‘high-risk’ score of 8 to 9 are recommended for mastectomy. More recently, Silverstein12 added patient age to the stratification scoring. A subsequent version of the model was simplified to margin width of >10 mm as a sole indicator of low risk for IBTR.13 Attempts at independent verification of these proposed risk stratification schemes have not consistently validated their utility.14–16 Despite this, the simplicity of the VNPI has enticed many clinicians to use it as a decision aid in the management of DCIS. In the current study, we used a modern cohort of patients subject to contemporary standards of mammographic, pathologic, and surgical processing and attempted to validate the clinical utility of the VNPI.

Table 1. Van Nuys Prognostic Index
  • VNPI indicates Van Nuys Prognostic Index.

  • *

    In Group 1 a VNPI of 3 to 4 was considered low risk, 5 to 7 was considered intermediate risk, and 8 to 9 was considered high risk.

  • In Group 2 a VNPI of 4 to 6 was considered low risk, 7 to 9 was considered intermediate risk and 10 to 12 was considered high risk.

VNPI (Group 1)*  
 Score123
  Size, mm≤1516–40>40
  Margin, mm≥101–9<1
  Grade123
VNPI (Group 2)  
 Score123
  Size, mm≤1516–40>40
  Margin, mm≥101–9<1
  Grade123
  Age, y>6040–60<40
VNPI (Group 3)
 RiskLow  Intermediate  High  
  Margin, mm≥101–9<1

MATERIALS AND METHODS

Data were retrospectively analyzed for 222 patients who were treated with excision alone for a mammographically detected DCIS at 4 institutions (Women and Infant's Hospital, Rhode Island Hospital, St. Elizabeth's Medical Center, and Tufts-New England Medical Center). The initial diagnosis of DCIS was established between 1987 and 2004, with the majority of cases accessioned between the years of 1995 and 2004 (Table 2). All the patients had a mammographically detected DCIS and were treated with a needle-localized excisional biopsy. The surgical specimen was measured and inked for orientation and margin assessment. Specimen radiographs were performed in all cases. After this procedure, serial sectioning was performed at 3 to 4-mm intervals along the long axis of the specimen (minimum number of sections, 15; mean number of sections, 26). When possible, disease distribution was mapped in 3 dimensions to accurately measure tumor size and excision margin width.

Table 2. Patient and Tumor Characteristics
  1. VNPI indicates Van Nuys Prognostic Index.

Age, y
 Mean57.0
 Range31–85
Year of diagnosis, % (no.)
 1987–199417.6% (39)
 1995–200482.4% (183)
Grade, % (no.)
 150.9% (113)
 232.9% (73)
 316.2% (36)
Tumor size, % (no.)
 ≤1.5 cm90.1% (200)
 1.6–4.0 cm8.6% (19)
 >4.0 cm1.4% (3)
VNPI Group 1, % (no.)
 Low (3–4)61.3% (136)
 Intermediate (5–7)36.5% (81)
 High (8–9)2.3% (5)
VNPI Group 2, % (no.)
 Low (4–6)71.6% (159)
 Intermediate (7–9)26.6% (59)
 High (10–12)1.8% (4)
VNPI Group 3: margin size, % (no.) 
 <1 mm14.0% (31)
 1–9.9 mm29.3% (65)
 ≥10 mm56.8% (126)

Multiple sections of each specimen were evaluated for the histopathologic subtype of DCIS, histologic and nuclear grade, maximum dimension of the lesion, and margin width. The nuclear grade of the DCIS was determined based on nuclear size, uniformity of size and shape, chromatin distribution, number of mitoses, and number and size of nucleoli.17

A second nuclear grading system, specifically proposed by Silverstein et al.18 for use in the VNPI, also was applied in a separate scoring of 200 patients in the cohort. In this system, nuclear grade and necrosis were used to determine the grade. The nuclear grade was based on the nuclear size relative to the dimension of an erythrocyte; high grade was >2 times the size of a erythrocyte, intermediate grade was 1 to 2 times the size, and low grade was <1 times the dimension of an erythrocyte. For calculation of the VNPI score, high grade was assigned 3 points, intermediate or low grade with necrosis received 2 points, and intermediate or low grade without necrosis was assigned 1 point.

Each patient was prospectively evaluated by a multidisciplinary management team that included review of the diagnostic images and excision specimen. Where appropriate, each patient was presented with treatment options including whole breast radiotherapy and/or tamoxifen. All the patients in this study cohort declined whole breast radiotherapy. For follow-up, patients were clinically examined every 3 to 6 months and received mammograms every 6 to 12 months. Suspected IBTR was confirmed by biopsy.

As an internal quality control measure, at the initiation of the study an independent pathologist reviewed a random sampling of 43% of the section slides to verify the accuracy of the original pathology reports. This review resulted in concordance of 100% of cases.

Patients were grouped in accordance with their VNPI scores as calculated by the various proposed VNPI schemes.11–13 Specifically, patients were first scored according to margin, lesion size, and both conventional and VNPI-specific nuclear grade and were referred to as VNPI Group 1. The entire cohort was then similarly reanalyzed with patient age as a fourth variable and referred to as VNPI Group 2. Lastly, all patients were assessed according to margin status alone and were referred to as VNPI Group 3.

Statistical Analysis

Freedom from IBTR for each risk group was calculated using the Kaplan-Meier method. Statistical significance between risk groups was calculated using the Wilcoxon test. Multivariate analyses were performed using Cox regression models to examine the independence and relative importance of different categoric variables.

RESULTS

The mean age of the patients was 57 years and the median follow-up was 4.6 years. An IBTR was detected in 8.6% of the patients and was evenly distributed between invasive and DCIS recurrences. Of the entire cohort of patients in this study, 50.9% had a grade 1 lesion, 90.1% had a lesion with maximal dimension of <15 mm, and 58.6% had a margin that was >10 mm (Table 2). Of the patients who did develop an IBTR, the mean age was 58.8 years, 73.7% of the patients had a lesion with a maximal dimension of <15 mm, 47.4% had a margin >10 mm, and 42.1% had grade 2 histology (Table 3).

Table 3. Proportional Distribution of IBTR According to Grade, Tumor Size, and VNPI Group
  1. IBTR indicates ipsilateral breast tumor recurrence; VNPI, Van Nuys Prognostic Index.

Grade, % (no.)
 136.8% (7)
 242.1% (8)
 321.1% (4)
Tumor size, % (no.)
 ≤1.5 cm73.7% (14)
 1.6–4.0 cm26.3% (5)
 >4.0 cm0.0% (0)
VNPI Group 1, % (no.)
 Low (3–4)31.6% (6)
 Intermediate (5–7)63.2% (12)
 High (8–9)5.3% (1)
VNPI Group 2, % (no.)
 Low (4–6)42.1% (8)
 Intermediate (7–9)57.9% (11)
 High (10–12)0% (0)
VNPI Group 3: margin size, % (no.) 
 Low (≥10 mm)47.4% (9)
 Intermediate (1–9.9 mm)26.3% (5)
 High (<1 mm)26.3% (5)

VNPI Group 1: Grade, Margin, and Size

The majority (61.3%) of patients had a VNPI score that placed them into the low-risk prognostic category. Only 5 patients were placed into the high-risk category (Table 2). At 5 years (approximate mean follow-up for the entire cohort), there were no significant differences noted among any of the risk-stratified groups with respect to rate of IBTR (Table 4). The data actuarially projected to 10 years demonstrated a trend with the low-risk group having the lowest rate of IBTR and the intermediate-risk group having the highest rate of IBTR (Fig. 1). Essentially identical results were observed when the scores were recalculated using the Van Nuys nuclear grading system (Table 4).

Figure 1.

Ipsilateral breast recurrence by Van Nuys Prognostic Index (VNPI) Group 1 categories.

Table 4. Freedom From IBTR
 5 year (95% CI)8 year (95% CI)Wilcoxon P
  1. IBTR indicates ipsilateral breast tumor recurrence; 95% CI, 95% confidence interval; VNPI, Van Nuys Prognostic Index; VN grading, Van Nuys nuclear grading system.

VNPI Group 1
 Low (3–4)96% (92–99)94% (88–99) 
 Intermediate (5–7)84% (74–94)75% (61–90) 
 High (8–9)100%75% (31–100) 
   .20
VNPI Group 1 (using VN grading)
 Low (3–4)95% (91–99)93% (87–98) 
 Intermediate (5–7)84% (72–95)79% (64–93) 
 High (8–9)75% (32–100)75% (32–100) 
   .48
VNPI Group 2
 Low (4–6)95% (91–99)93% (88–98) 
 Intermediate (7–9)83% (72–95)69% (51–87) 
 High (10–12)100%100% 
   .19
VNPI Group 2 (using VN grading)   
 Low (4–6)94% (90–98)93% (87–98) 
 Intermediate (7–9)84% (71–96)73% (56–91) 
 High (10–12)100%100% 
   .39
VNPI Group 3
 Low (≥10 mm)92% (86–97)85% (70–99) 
 Inter (1–9.9 mm)91% (82–100)87% (76–99) 
 High (<1 mm)91% (78–100)78% (59–98) 
   .98

VNPI Group 2: Grade, Margin, Size, and Patient Age

The majority of patients (71.6%) were categorized as low risk when age was added as a variable (Table 2). The majority of IBTR cases were noted in the intermediate-risk category (52.6%). The percentage of subjects with no IBTR at 5 years was 95% and was 83% for the low-risk and intermediate-risk groups, respectively (P = .19) (Table 4) (Fig. 2). Similar results were observed when the Van Nuys nuclear grading system was substituted for the conventional grading system.

Figure 2.

Ipsilateral breast recurrence by Van Nuys Prognostic Index (VNPI) Group 2 categories.

VNPI Group 3: Margin Status Alone

The majority of patients in the study cohort had a margin of ≥10 mm (56.8%), whereas 14.0% of patients had a margin <1 mm (Table 2). Of the patients who developed an IBTR, 57.6% had a margin of ≥10 mm. There was no significant difference noted in the 5-year rate of IBTR for patients with a margin <1 mm, those with a margin of 1 to 9.9 mm, or those with a margin ≥10 mm (91%, 91%, and 92%, respectively) (Table 4). When data were actuarially projected to 10 years, a margin <1 mm was found to be associated with the greatest risk of IBTR (Fig. 3).

Figure 3.

Ipsilateral breast recurrence by Van Nuys Prognostic Index (VNPI) Group 3 categories.

Tamoxifen

Tamoxifen was administered to 31.4% of the cohort. The estimated 5-year IBTR-free survival for patients who received tamoxifen was 95% and was 94% for those who did not (P = 1.0).

DISCUSSION

There have been several prospectively randomized trials evaluating the role of radiotherapy after lumpectomy for DCIS. Among the most prominent, the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-17 study enrolled 818 patients and compared lumpectomy alone with lumpectomy and whole breast irradiation.19 At a median follow-up of >10 years, breast irradiation was associated with a reduction in the rate of invasive IBTR from 17% to 8% and that of noninvasive IBTR was reduced from 15% to 8%. The European Organization for Research and Treatment of Cancer (EORTC) 10853 trial used an identical randomization scheme and found that irradiation was associated with a reduction in both invasive and DCIS IBTR of approximately 50%.20

Despite the consistent results of large prospective clinical trials that support the use of breast radiotherapy, a pressing clinical question remains as to which patients with DCIS might be reasonably considered for excision alone. The VNPI was proposed as an attempt to incorporate several variables generally considered to influence the relative risk of IBTR and assign a prognostic category. Understandably, the VNPI has generated broad interest because it is an appealingly simple solution to what can sometimes be complex clinical management decisions. However, its routine application must be tempered by the finding that it is constructed on a foundation of data derived from a single source that has been analyzed retrospectively. There have been a limited number of independent attempts to validate the VNPI based on data collected both retrospectively15, 16 and prospectively.14 In the current series, we used data accumulated in the contemporary practice of multidisciplinary breast cancer management. As such, we attempted to validate the VNPI with information regarding tumor size, grade, margin status, and patient age that are typically presented to a multidisciplinary breast tumor review conference at both academic and community-based medical centers. To our knowledge, the current study of 222 patients who underwent excision alone for DCIS is one of the largest of its kind published to date. All the patients had adequate clinical and pathologic information to score in accordance with the VNPI risk categories.

The results of the current study found no statistically significant difference in the risk of recurrence at 5 years among the prognostic groups as stratified by the VNPI. This finding held true regardless of how the prognostic score was assigned. However, it must be emphasized that, due to the very small sample size associated with the high-risk group, this study is best considered a comparison limited to VNPI of low-risk and intermediate-risk categories.

We have limited our conclusions regarding the validation of the VNPI to the 5-year follow-up point becaue this approximates the median follow-up duration of the current study cohort. It is interesting to note that actuarial projections beyond 5 years did demonstrate a trend toward a separation of the low-risk and intermediate-risk categories. The possibility that, with additional follow-up, a statistically significant difference may emerge cannot be eliminated.

In addition to the multifactorial VNPI prognostic scoring, the current study also evaluated margin width as the sole determinant of IBTR risk. In contrast to the findings of Silverstein et al.,13 we found no significant difference in the 5-year rate of IBTR for patients categorized by a margin width of <1 mm, 1 mm to 9.9 mm, or >10 mm. Again, however, actuarial projections do suggest a trend toward an increased risk of IBTR in cases with a <1-mm margin and additional follow-up may lead to the emergence of statistically significant differences.

The specimen processing in the current study was rigorous and reflective of contemporary practices that are in widespread use at both community and academic centers. Nonetheless, it is possible that these practices do not fully reflect the processing techniques used in the Van Nuys experience11–13, 18 that may influence any attempt to validate the VNPI.

A weakness in our retrospective study is the potential confounding effect of tamoxifen. However, data from the current study did not find a statistically significant difference in the rate of IBTR for patients who did versus those who did not receive tamoxifen.

A major difficulty in a retrospective study of DCIS is establishing and verifying a uniform assessment of grade and margin width. We attempted to address this problem by performing an internal quality control process by which an independent pathologist reviewed a random sample of 43% of the specimen slides. We also attempted to address the potential inconsistency inherent in assigning a nuclear grade by analyzing the data with 2 distinct grading systems. However, the consistent and accurate assessment of margin status remains a challenge and occasionally can be difficult to ascertain with certainty. For example, in the patient who has undergone a reexcision with the finding of no residual tumor, it is often impossible to verify with certainty that the additional tissue was taken from the appropriate location unless the specimen has been meticulously marked and oriented.

It also should be noted that the overall rate of IBTR at 5 years in the current study was quite low—certainly lower than that reported in the original VNPI articles of Silverstein et al.11–13, 18 or that noted in the randomized studies of the National Surgical Adjuvant Breast and Bowel Project (NSABP)19 and EORTC.20 This can be accounted for in part by the relatively limited duration of follow-up and the disproportionate distribution of our study population toward low-grade histology. However, this low recurrence rate also may reflect the contemporary clinical practice surrounding the improved management of DCIS with the widespread use of thorough mammographic evaluation, modern surgical techniques, and disciplined specimen processing to ensure adequate margin evaluation.

Conclusions

The findings of the current study suggest that VNPI or margin width alone is not a valid tool to assist in the stratification of patients after excision alone for the risk of IBTR at 5 years. Further follow-up may strengthen the predictive utility of the various VNPI classification schemes.

Ancillary