SEARCH

SEARCH BY CITATION

Keywords:

  • Wilms tumor;
  • nuclear unrest;
  • anaplasia;
  • favorable histology;
  • p53;
  • immunohistochemistry;
  • kidney;
  • pediatric;
  • pathology

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

BACKGROUND

Nuclear unrest is a term applied to Wilms tumors (WT) with favorable histology that show nuclear enlargement similar to that seen in anaplasia but with no abnormal mitotic figures. This study was undertaken to evaluate the clinical and pathogenetic significance of WT with nuclear unrest.

METHODS

The authors reviewed primary nephrectomy specimens and clinical data from 173 patients who were treated for WT on 1 of 5 consecutive clinical trials at St. Jude Children's Research Hospital. A subset of 70 patients was selected for p53 immunohistochemistry. Seventeen samples of recurrent tumors were also reviewed.

RESULTS

WT with nuclear unrest had age and stage distribution similar to that for WT with favorable histology. Patients who had tumors with favorable histology, nuclear unrest, and anaplastic histology had 5-year cumulative incidences of recurrence of 12.2%, 22.2%, and 36.4%, respectively (P = 0.010). Despite the difference in recurrence incidence, patients who had tumors with nuclear unrest and patients who had tumors with favorable histology had nearly identical overall survival. Immunostaining for p53 was positive in 2 of 24 tumors (8.3%) with favorable histology, 1 of 25 tumors (4%) with nuclear unrest, and 16 of 21 tumors (76%) with anaplastic histology (P < 0.001). Histologic review of samples from recurrent tumors showed that 7 of 10 of primary tumors converted from favorable histology to either nuclear unrest or anaplastic histology in the recurrent tumor.

CONCLUSIONS

WT with nuclear unrest more closely resembles favorable histology than anaplastic histology, both clinically and pathogenetically. Analysis of samples from recurrent tumors suggests that WT with nuclear unrest represents an intermediary in the continuum from favorable histology to anaplastic histology. For treatment stratification, it is appropriate to continue grouping nuclear unrest with favorable histology, although a larger analysis will be necessary to confirm the current findings. Cancer 2003;97:2318–26. © 2003 American Cancer Society.

DOI 10.1002/cncr.11325

Histologic classification is the most important prognostic indicator for patients with Wilms tumor. Traditionally, tumors are divided into two groups, favorable and anaplastic histologies. In previous National Wilms Tumor Studies (NWTS), tumors in the anaplastic group comprised approximately 5% of Wilms tumors.1 Despite remarkable success in the treatment of patients who have Wilms tumor with favorable histology, the estimated recurrence free survival rate for patients with anaplastic tumors is only 40–50%, even with the use of intensive chemotherapy and abdominal irradiation.2–4

Anaplastic Wilms tumors are characterized by tumor cells with significant nuclear enlargement; hyperchromatism; and atypical, multipolar mitotic figures. The nuclear enlargement and hyperchromatism correlate with increased cellular DNA content;5, 6 and the atypical, multipolar mitotic figures presumably represent markers of increased genetic mutability with multiple mitotic errors.3 Several studies have shown that detection of p53 by immunostaining, with or without demonstrated mutations, is associated strongly with anaplasia in Wilms tumor, whereas p53 staining is uncommon in tumors with favorable histology.7–14 Because a functional p53 protein has been linked to cell-cycle checkpoint control,15 the presence of p53 mutations may contribute to the increased DNA content and the irregular mitotic figures observed in anaplastic Wilms tumor. Moreover, the importance of p53 in directing cells with DNA damage into an apoptotic pathway may provide a molecular basis for the insensitivity of anaplastic Wilms tumor to therapeutic intervention.2

Some Wilms tumors with favorable histology show disturbing nuclear enlargement, cytologic atypia, and histologic disarray known as nuclear unrest. Wilms tumors with nuclear unrest contain tumor cells with enlarged, hyperchromatic nuclei but do not have the enlarged, multipolar mitotic figures required to meet the criteria for anaplasia.3, 5 Patients with tumors that contain nuclear unrest currently are treated on regimens for patients with tumors that have favorable histology; however, the cytologic similarities between tumors with nuclear unrest and anaplastic tumors provide a source of anxiety for pathologists and treating clinicians alike. Because the clinical significance of nuclear unrest in tumors with otherwise favorable histology has not been studied systematically, the appropriate treatment for patients with such tumors is unclear.

To evaluate the clinical significance of nuclear unrest, a retrospective analysis of 173 patients with Wilms tumors was conducted. In addition, to further elucidate the molecular genetic etiology of nuclear unrest, p53 immunohistochemistry was performed to determine whether these tumors have the strong staining pattern that characterizes most anaplastic Wilms tumors.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Patients

Pathologic material and clinical information from patients who were treated on one of five consecutive clinical trials at St. Jude Children's Research Hospital between 1973 and 1998 were collected. One hundred seventy-three patients had undergone a primary prechemotherapy nephrectomy or biopsy and had adequate archival tissue for review. Clinical characteristics and outcome data were available for all patients. The treatment regimens and staging system have been described elsewhere.16 This retrospective study was approved by the St. Jude Institutional Review Board.

Histologic Classification

The tumors were reviewed and classified according to standard NWTS Group criteria for favorable histology and anaplasia (Fig. 1A,B). A diagnosis of anaplasia was made when the following three criteria were present: 1) an increase in greatest nuclear dimension at least three times that of adjacent nuclei of the same cell type; 2) hyperchromatism of the enlarged nuclei; and 3) the presence of abnormal, enlarged, usually multipolar mitotic figures.3 For this study, focal and diffuse anaplasia were considered as a single group. Within the favorable histology group, the criteria proposed by Zuppan et al. were used to define nuclear unrest.3 Those authors graded the severity of nuclear unrest according to a three-tiered scheme: Grade 1 unrest reflects minimal disorder, and Grade 3 unrest reflects the nuclear atypia and enlargement similar to that seen in anaplastic Wilms tumors but without the associated enlarged, multipolar mitotic figures (Fig. 1C). Grade 2 nuclear unrest is intermediate between Grades 1 and 3. Given the inherent subjectivity in these criteria, grades of nuclear unrest were not assigned, and all tumors with nuclear unrest, regardless of grade, were grouped together for the purposes of this study. The pathologist conducting this review (D.A.H.) was blinded to the clinical characteristics and patient outcomes.

thumbnail image

Figure 1. (A) Wilms tumor with favorable histology. (B) An anaplastic Wilms tumor. Note the enlarged nuclei and the tetrapolar mitotic figure. (C) Wilms tumor with severe nuclear unrest (hematoxylin and eosin staining; original magnification × 200).

Download figure to PowerPoint

Immunohistochemistry

Seventy of the original 173 tumors, including 21 anaplastic tumors, 25 tumors with nuclear unrest, and 24 tumors with favorable histology, were selected for a study of p53 expression using immunohistochemistry. One to three blocks containing tumor cells representative of the histologic classification from each tumor were selected for staining. Immunohistochemical analysis was performed on 4-μm-thick, formalin fixed, paraffin embedded (FFPE) tissue sections. Tissue sections prepared for antibody application were pretreated with heat-induced epitope retrieval performed in a steamer using ethylenediamine tetraacetic acid buffer, pH 8.0, for 30 minutes. The mouse monoclonal antihuman p53 antibody (DO-7; Dako, Carpinteria, CA) at a 1:50 dilution was applied and incubated for 32 minutes. DO-7 is a commonly used antibody that recognizes both wild-type and mutant p53. Slides were then developed with a modified avidin-biotin-peroxidase complex technique17 using an autostainer (ES; Ventana, Tucson, AZ). The assay was optimized previously using FFPE sections from known p53 mutated breast carcinoma and p53 mutated and wild-type pediatric adrenocortical carcinoma. Staining was scored by three reviewers (D.A.H., T.D.S., and P.K.S.) and was based on distribution (focal: staining of a single area on the slide; multifocal: multiple, discrete areas of staining on the slide; and diffuse: widespread staining) and strength using a 4-point scale (0, none; 1 +, weak; 2 +, moderate; and 3 +, strong). The percentages of positive cells were measured by crude visual estimate and recorded. Tumors with crisp nuclear staining of moderate and strong strength in > 20% of cells were considered positive. The 20% cut-off value was chosen arbitrarily and is consistent with that used in other studies of p53 immunohistochemistry.18

Statistical Methods

Associations among variables were examined using a number of statistical tests: The Fisher exact test, the chi-square test, and the Kruskall–Wallis test were used to assess correlations between categorical (nominal or ordered) variables. The Wilcoxon–Mann–Whitney test was used to examine age at diagnosis (as a continuous variable) by histologic group. Exact tests were used when an expected sample size was less than five. The cumulative incidence (CI) of disease recurrence/progression (measured from the date of diagnosis to the date of first documented recurrence or progression) was estimated by the methods of Kalbfleisch and Prentice.19 The Gray test20 was used to compare the CI of recurrence/progression among groups. Using the Gray test enabled us to adjust for the competing risks of death or second malignancy prior to disease recurrence or progression. Pair-wise comparisons were made only if the overall P value was < 0.05. Because, currently, there are no available statistical methods for determining statistical power using the Gray test, power calculations were not performed. However, we estimate that the power to determine differences in the CI of recurrence/progression was low. The CI of death also was estimated, and the Gray test was used to compare CI distributions among factors. No adjustments were made for multiple comparisons given the exploratory nature of this study.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Patient Characteristics

Table 1 summarizes the clinical characteristics of the 173 patients who were included in the outcome analysis as well as associations between clinical characteristics and histologic subtype. Overall, 71.7% of tumors were classified with favorable histology, 15.6% tumors were classified with nuclear unrest, and 12.7% of tumors were classified with anaplastic histology. Patient age at the time of diagnosis differed according to histologic subtype (P = 0.015). Patients with anaplastic tumors were more likely to present at an older age compared with patients who had tumors with favorable histology (P = 0.004) or nuclear unrest (P = 0.015). There was no difference in age distribution between patients who had tumors with favorable histology and patients who had tumors with nuclear unrest (P = 0.95). Tumor stage also differed according to histologic group (P = 0.020). Patients with anaplastic tumors tended to present with higher stage disease compared with patients who had tumors with favorable histology (P = 0.061) or nuclear unrest (P = 0.008). No difference in stage distribution was observed between patients who had tumors with favorable histology and patients who had tumors with nuclear unrest (P = 0.184). No differences were observed for race or gender by histologic subtype.

Table 1. Patient Characteristics
CharacteristicNo. of patients (%)Favorable (n = 124 patients)Nuclear unrest (n = 27 patients)Anaplastic (n = 22 patients)P value
  1. NWTSG: National Wilms Tumor Study Group staging system.

Stage (NWTSG)     
 I 39 (23)2775 
 II 44 (25)33101 
 III 51 (30)39480.26
 IV 30 (17)1848 
 V  9 (5)720 
 I/II 83601760.020
 III/IV 8157816 
Gender     
 Female105 (61)7714140.59
 Male 68 (39)47138 
Race     
 White130 (75)9816160.094
 African American 43 (25)26116 
Age at diagnosis (yrs)     
 Median3.73.43.35.20.015
 Range0.2–16.80.2–14.80.8–9.80.9–16.8 

Outcome Analysis

Among 173 patients studied, 149 patients (86.1%) were alive with a median follow-up of 15.4 years (range, 1.4–27.7 years); 115 survivors (77.2%) had been contacted within 1 year of the analysis date. First events consisted of recurrence or progressive disease in 29 patients, second malignancies in 3 patients, and death prior to any other events in 2 patients. One of these patients died as a result of surgical complications after undergoing second-look surgery, and the other patient died suddenly without a clear etiology 6 months after completion of therapy for Wilms tumor. The median time from diagnosis to recurrence or progression was 7.8 months and ranged from 19 days to 3.4 years. For all patients, the 5-year estimate of the CI of recurrence was 16.8% ± 2.9%, and the 5-year CI of death was 12.8% ± 2.6%.

Table 2 summarizes prognostic factors for recurrence and death. Age was categorized above or below the median age of all patients, 3.7 years. Age > 3.7 years at the time of diagnosis and advanced tumor stage (Stage III and IV) were associated with higher CI of both recurrence and death (Table 2). There was a difference in the CI of recurrence according to histologic subtype (P = 0.010): Patients in the favorable histology group had a 5-year CI of recurrence of 12.2% ± 3.0%, whereas patients in the anaplastic group had a 5-year CI of recurrence of 36.4 ± 10.6%. Patients in the nuclear unrest group had an intermediate 5-year CI of recurrence of 22.2% ± 8.2% (Fig. 2). In pairwise comparisons, the CI of recurrence differed between patients in the favorable histology group and patients in the anaplastic group (P = 0.003) and did not differ between patients in the nuclear unrest group and patients in the favorable histology group (P = 0.17) or between patients in the nuclear unrest group and patients in the anaplastic group (P = 0.22). Although methods for calculating power are not available for the statistical test that we used (Gray test), we estimate that the statistical power was low for comparing the CI of recurrence between the nuclear unrest group and the other histologic groups.

Table 2. Factors Prognostic of Recurrence or Disease Progression and Death in Patients with Wilms Tumor
FactorNo. of patientsRecurrence or progressionDeath
5-yr CI estimate (1 SE)%P value5-yr CI estimate (1 SE)%P value
  1. CI: cumulative incidence; SE: standard error.

Age at diagnosis (yrs)     
 ≤ 3.78810.3 (3.43)0.027 6.9 (2.7)0.017
 > 3.78523.6 (4.6) 19.0 (4.3) 
Gender     
 Female10516.3 (3.6)0.8314.4 (3.5)0.30
 Male6817.7 (4.7) 10.3 (3.7) 
Race     
 Caucasian13015.5 (3.2)0.3812.4 (2.9)0.97
 African-American4320.9 (6.3) 14.0 (5.4) 
Stage     
 I, II8310.8 (3.4)0.052 4.9 (2.4)< 0.001
 III, IV, V9022.4 (4.5) 20.2 (4.3) 
 I, II8310.8 (3.4)0.084 4.9 (2.4)< 0.001
 III, IV8121.0 (4.6) 20.0 (4.5) 
Histology     
 Favorable12412.2 (3.0)  9.0 (2.6) 
 Nuclear unrest2722.2 (8.2)0.01011.1 (6.2)0.003
 Anaplastic2236.4 (10.6) 36.4 (10.6) 
thumbnail image

Figure 2. Cumulative incidence of recurrence or disease progression in patients with Wilms tumor by histologic subgroup. FH: favorable histology; NU: nuclear unrest.

Download figure to PowerPoint

In the analysis of death, the marked difference in outcome according to histology persisted (P = 0.003): Patients in the favorable histology group had a 5-year CI of death of 9.0% ± 2.6%, whereas patients in the anaplastic group had a 5-year CI of death of 36.4% ± 10.6%. It is interesting to note that the 5-year CI of death for patients in the nuclear unrest group was similar to that for patients in the favorable histology group (11.1% ± 6.2%) (Fig. 3). In pairwise comparisons, the CI of death differed between patients in the favorable histology group and patients in the anaplastic group (P = 0.001) and between patients in the nuclear unrest group and patients in the anaplastic group (P = 0.035), but not between patients in the nuclear unrest group and patients in the favorable histology group (P = 0.92). Neither gender nor race was prognostic of recurrence or death (Table 2).

thumbnail image

Figure 3. Cumulative incidence of death in patients with Wilms tumor by histologic subgroup. FH: favorable histology; NU: nuclear unrest.

Download figure to PowerPoint

p53 Immunohistochemistry Analysis

Seventy tumors were selected for p53 immunohistochemical analysis, representing a nearly equal distribution of tumors with favorable histology, anaplasia, and nuclear unrest. A summary of the immunohistochemistry results is provided in Table 3. There was 100% concordance between reviewers for absolute positive and negative p53 calls. Sixteen of 21 anaplastic tumors (76.2%) were positive for p53 expression, all of which had a strong staining (3 +) pattern. Eleven of those 16 tumors showed p53 expression in a diffuse pattern with > 90% positive cells (Fig. 4A). Staining for p53 was seen in all three cell types: blastema, epithelium, and stroma (Fig. 4B,C). The majority of tumors with favorable histology and nuclear unrest showed only rare, scattered tumor cells with nuclear positivity for p53 (Fig. 5). Only 2 of 24 tumors with favorable histology (8.3%) and 1 of 25 tumors with nuclear unrest (4.0%) met the criteria for positivity, with > 20% of nuclei showing moderate (2 +) staining for p53. In some tumors with favorable histology and nuclear unrest that were classified as p53 negative, weak staining for p53 was seen in small groups of cells surrounding areas of necrosis, a finding that may be consistent with wild type p53 expression in a region of hypoxia.21

Table 3. Staining and Histology for p53
Histologic typeNo. of patients (%)
p53 positivep53 negativeTotal
Favorable 2 (8%)2224
Nuclear unrest 1 (4%)2425
Anaplasia16 (76%)521
Total195170
thumbnail image

Figure 4. (A) Typical diffuse pattern of strong p53 staining in anaplastic Wilms tumor. (B,C) p53 was seen in blastema (B), epithelium (C), and stromal cell types. (D) High-power photomicrograph (anti-p53 immunostaining with hematoxylin counterstain; original magnification × 100 in A; × 200 in B, C; × 400 in D).

Download figure to PowerPoint

thumbnail image

Figure 5. Negative p53 staining in tumors with favorable histology and nuclear unrest. Rare tumor cell nuclei showed weak positivity, but most were negative (anti-p53 immunostaining with hematoxylin counterstain; original magnification × 200).

Download figure to PowerPoint

Analysis of Recurrent Tumor Samples

Seventeen of 29 patients who developed recurrent disease had archival material available from both the diagnostic specimen and the recurrent specimen. Among patients who had tumors with favorable histology at diagnosis, three patients had tumors with favorable histology, three patients had tumors with nuclear unrest, and three patients had anaplastic tumors at recurrence (Table 4). One additional patient had a tumor with favorable histology at first recurrence and a tumor with nuclear unrest at second recurrence. Among five patients who had tumors with nuclear unrest at diagnosis, four patients had tumors with nuclear unrest and one had an anaplastic tumor at recurrence. Two patients who had tumors with anaplastic histology at diagnosis had concordant histology at recurrence. No patients who had tumors with nuclear unrest or anaplastic histology had tumors with favorable histology at recurrence. All recurrent tumors with anaplasia were p53 positive by immunostaining, whereas recurrent tumors with favorable histology and nuclear unrest were p53 negative.

Table 4. Histologic Classification of Primary and Recurrent Wilms Tumors
Primary histologyRecurrent histologyStaining for p53Recurrence location
Primary tumorRecurrent tumor
  1. FH: favorable histology; NA: not available; −: negative; NU: nuclear unrest; 2nd: second recurrence; AH: anaplastic histology; +: positive; LN: lymph node.

FHFHNALung
FHFHNAOvary
FHFHNANALung
FHNUNALung
FHNUNALung
FHNUNALung
FHFH, NU (2nd)NA−, − (2nd)Bone, lung (2nd)
FHAH+Lung
FHAHNAKidney
FHAH+Peritoneum
NUNULung
NUNURenal fossa
NUNULung
NUNUNALung
NUAH+Lung
AHAH++Axillary LN
AHAH++Lung

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Wilms tumor with nuclear unrest has morphologic features that place it between tumors with favorable histology and anaplastic tumors. The significance of this limited morphologic change is unclear, and it is uncertain whether patients who have these tumors warrant more aggressive treatment than the treatment provided commonly for patients who have tumors with favorable histology. We observed that Wilms tumors with nuclear unrest more closely resembled tumors with favorable histology than tumors with anaplastic histology with regard to the clinical features of stage and patient age at presentation. Moreover, there was no difference in overall survival between patients who had tumors with nuclear unrest and patients who had tumors with favorable histology, whereas patients who had tumors with anaplastic histology had a significantly lowershorter estimated survival. Taken together, our observations suggest that it is appropriate to continue treating patients who have tumors with nuclear unrest the same as patients who have tumors with favorable histology.

A caveat to this recommendation is that there may be a difference in the recurrence rate between patients who have tumors with favorable histology and patients who have tumors with nuclear unrest. Patients in the nuclear unrest group had a CI of recurrence between that of patients in the favorable histology group and patients in the anaplastic histology group, although the differences were not statistically significant. Our study lacked statistical power to discern differences in recurrence incidence; therefore, analysis of more patients will be necessary to confirm whether nuclear unrest predicts recurrence.

Wilms tumors with nuclear unrest were biologically similar to tumors with favorable histology, in that p53 overexpression was observed infrequently in these histologic subtypes, whereas p53 overexpression was observed commonly in anaplastic tumors. Because p53 overexpression often reflects p53 mutation, the current study data suggest that p53 mutation is uncommon in Wilms tumor with nuclear unrest. It is likely that functional p53 protein in tumors with nuclear unrest accounts for their continued sensitivity to therapy at recurrence.

The pathogenetic correlations between favorable histology, nuclear unrest, and anaplasia have not been established. Several lines of evidence suggest that anaplastic Wilms tumors arise from the accumulation of genetic damage in Wilms tumors with favorable histology. First, focal anaplasia, as an entity, is defined as anaplastic nuclear changes confined to one or more clearly defined loci and is consistent with a clonal genetic event in a tumor with favorable histology.5 Second, some bilateral or multifocal tumors within the same patient have discordant histology.22 Although these multifocal tumors presumptively share the same predisposing genetic event, the fact that a subset of tumors becomes anaplastic suggests that additional genetic mutations have occurred. Third, microdissection analyses of tumors containing both anaplastic histology and favorable histology showed that p53 mutations were restricted to areas of anaplasia.8 This supports the notion that the acquisition of p53 mutations is inherent to the process of anaplastic progression. Given what has been learned about the multistep progression of tumorigenesis in many epithelial tumors in adults, the involvement of p53 at a later stage in tumor development is not surprising, although this is an uncommon phenomenon in most other pediatric tumors.23 Finally, patients with anaplastic Wilms tumors have an older age at the time of presentation compared with their counterparts who have tumors with favorable histology, and anaplastic Wilms tumors are seen rarely in children age < 2 years.3 The older patient age at presentation suggests that it takes time for the additional genetic insults that lead to anaplasia to accumulate.

We propose that nuclear unrest represents an intermediary in the spectrum from favorable histology to anaplastic histology. This premise is supported by our observation that four patients who had tumors with favorable histology at the time of diagnosis had tumors with nuclear unrest at the time of recurrence, and one patient who had a tumor with nuclear unrest at the time of diagnosis had an anaplastic tumor at the time of recurrence. The lack of p53 overexpression in tumors with nuclear unrest indicates that p53 dysfunction is not responsible for the nuclear enlargement seen with nuclear unrest, but these tumors subsequently may incur p53 mutations and convert to bona fide anaplasia (Fig. 6). We identified three patients who had primary tumors with favorable histology at the time of diagnosis and anaplastic tumors at the time of recurrence, suggesting that favorable histology also may convert directly to anaplasia without going through a nuclear unrest intermediary. It is possible, however, that the nuclear unrest phase was undetected before the recurrence was recognized. We cannot exclude the possibility that nuclear unrest represents a variant of favorable histology that lies outside the pathogenetic lineage to anaplasia. Gene expression studies may clarify further the correlations between the histologic subtypes of Wilms tumor.

thumbnail image

Figure 6. Proposed model of pathogenetic correlations between Wilms tumor histologic subtypes. A review of samples from recurrent Wilms tumors favors a model in which tumors with favorable histology (FH) undergo mutation(s) that give rise to nuclear unrest (NU), which may then convert to anaplasia on inactivation of p53. Alternatively, tumors with favorable histology may convert directly to anaplastic histology (AH).

Download figure to PowerPoint

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The authors thank Charlene Henry, Nettie Turner, and Inez Lewis for their immunohistochemistry support and Angela Baker, Alice Slusher, and Carolyn Beard for their histology expertise. They also thank Dr. Daniel Green for his insightful comments on the article, Angie Moore for administrative help, and Dr. James Shear for editorial advice.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES