Wilms tumor is the most common childhood renal tumor. Although the majority of patients with favorable histology Wilms tumor (FHWT) have good outcomes, some patients still experience disease recurrence and death from disease. The goal of the current study was to determine whether tumor-specific chromosome 1q gain is associated with event-free survival (EFS) and overall survival (OS) in patients with FHWT.
Unilateral FHWT samples were obtained from patients enrolled on National Wilms Tumor Study-4 and Pediatric Oncology Group Wilms Biology Study (POG 9046). 1q gain, 1p loss, and 16q loss were determined using multiplex ligation-dependent probe amplification.
The 8-year EFS rate was 87% (95% confidence interval [95% CI], 82%-91%) for the entire cohort of 212 patients. Tumors from 58 of 212 patients (27%) displayed 1q gain. A strong relationship between 1q gain and 1p/16q loss was observed. The 8-year EFS rate was 76% (95% CI, 63%-85%) for patients with 1q gain and 93% (95% CI, 87%-96%) for those lacking 1q gain (P = .0024). The 8-year OS rate was 89% (95% CI, 78%-95%) for those with 1q gain and 98% (95% CI, 94%-99%) for those lacking 1q gain (P = .0075). Gain of 1q was not found to correlate with disease stage (P = .16). After stratification for stage of disease, 1q gain was associated with a significantly increased risk of disease recurrence (risk ratio estimate: 2.72; P = .0089).
Wilms tumor is the most common primary renal tumor occurring in childhood, with approximately 500 cases identified each year in the United States. The majority of Wilms tumors occur in children aged < 5 years, and are characterized by a triphasic histologic pattern consisting of blastemal, stromal, and epithelial elements. A small minority of Wilms tumors contain anaplastic histology consisting of nuclear enlargement, nuclear atypia, and irregular mitotic figures, which is associated with an increased risk of disease recurrence.[2, 3] The absence of anaplastic features identifies patients with favorable histology Wilms tumor (FHWT), which represents the patient population analyzed in the current study.
Multidisciplinary collaboration and research through the National Wilms Tumor Study Group (NWTSG), now the Renal Tumor Committee of the Children's Oncology Group (COG), as well as the European pediatric cooperative groups have led to dramatic improvements in survival for the majority of patients with FHWT.[4-10] Multiple patient and tumor factors have been identified as being prognostically significant in the NWTSG/COG patient cohorts, including tumor histology, disease stage, tumor-specific loss of heterozygosity (LOH) for chromosomes 1p and 16q,[10, 12] and patient age and tumor weight.[13-15] Although the majority of patients with FHWT have good outcomes, many patients still experience disease recurrence and death from disease, even among those patients with lower disease stages. In addition, currently available therapeutic approaches expose patients to significant risk for both immediate and late morbidity and mortality, including cardiac and hepatic toxicity; secondary malignancies; and pregnancy complications. Additional prognostic factors are necessary to prospectively identify those patients at the time of diagnosis who are at greater risk of disease recurrence. LOH of 1p and 16q in patients with Wilms tumor was first described from a cohort of NWTSG patients enrolled on the third and fourth National Wilms tumor studies (NWTS-3 and NWTS-4), which demonstrated inferior outcomes for patients with LOH for 16q, and trends toward inferior outcomes for patients with LOH for 1p. Furthermore, work using prospectively gathered samples as part of the NWTS-5 provided compelling data that patients with combined LOH for 1p and 16q had inferior event-free survival (EFS) and overall survival (OS), regardless of stage of disease. However, only 4.6% of patients with FHWT (76 of 1656 patients) had tumors with combined LOH 1p/16q, and combined LOH for 1p/16q was found to be present in only 9.4% of cases of disease recurrences (20 of 213 recurrences).
Several additional studies have identified other genetic changes in FHWT that are associated with outcome. The change that has been consistently reported in all such studies is a gain of chromosome 1q.[20-31] Despite the strength of the data previously reported, all these studies were performed in convenience samples that were not treated consistently. Furthermore, to the best of our knowledge, limited data have been published demonstrating the relationship between 1q gain and outcome within different tumor stages. It is important to note that 1q gain and LOH at 1p/16q are not independent events. LOH at 1p and 16q often arises through chromosomal translocations that also result in 1q gain. The goal of the current study was to determine whether 1q gain, analyzed using multiplex ligation-dependent probe amplification (MLPA), is associated with EFS and OS in a carefully stratified cohort of patients with FHWT enrolled on NWTS-4. The study also attempted to better define the relationship between 1q gain and 1p/16q LOH.
MATERIALS AND METHODS
Patients included in the current study were drawn from those who registered prospectively after providing informed consent on the Pediatric Oncology Group Wilms Biology Study (POG 9046). Although some of the POG 9046 patients were enrolled on NWTS-3, the entire cohort from this study came from patients enrolled on NWTS-4, which represents a slightly different cohort from that reported previously. Requirements for inclusion into the current study were the presence of unilateral renal disease, FHWT confirmed by central pathology review, and registration with full eligibility and follow-up on NWTS-4.
Multiplex Ligation-Dependent Probe Amplification
MLPA was performed as previously described using a synthetic probe mixture and minor modifications. Gene-specific left and right probes were created using AlleleID 7.70 (Premier Biosoft International, Palo Alto, Calif). Probes for 1q were selected to include the minimum region of gain in 1q22 to 1q23 that was previously reported.[25, 30] All probe and primer target-specific sequences are provided with controls in Table 1. Universal forward and reverse sequences were attached to the left and right probes, respectively. The probe sizes were increased using random sequences to obtain a specific amplicon size. The right probe was modified with the addition of 5′ phosphorylation. The probes were synthesized by Integrated DNA Technologies (Coralville, Iowa). A probe master mix contained a final concentration of 2.6 nM for each left probe and 2.6 nM for each right probe. Approximately 250 ng of DNA in a volume of 5 μL was denatured at 98°C for 5 minutes. A mixture of 1.5 μL of MLPA buffer (1.5 M of KCl, 300 mM Tris-HCl [pH 8.5], and 1 mM of ethylenediaminetetraacetic acid) and 1.5 μL of a probe set (6.9 pM for each hemiprobe) was added. The mixture was heated for 1 minute at 95°C followed by 16 hours to 18 hours at 60°C to allow the MLPA hemiprobes to hybridize. Next, 32 μL of ligase-65 mixture (dilution buffer containing 2.6 mM of MgCl2, 5 mM of Tris-HCl [pH 8.5], 0.013% nonionic detergents, 0.2 mM of nicotinamide adenine dinucleotide, and 1 U of ligase-65 enzyme) was added to each sample for ligation of hybridized hemiprobes during 15 to 20 minutes of incubation at 54°C, followed by 5 minutes of incubation at 98°C to inactivate the ligase. The amplification step was performed in a 25-μL reaction volume using 2.5 μL of 10X polymerase chain reaction (PCR) Gold buffer, 4 μL of 25 mM MgCl2, 0.25 μL of AmpliTaq Gold (5U/μL) (Applied Biosystems, Foster City, Calif), 0.5 μL of dNTP, 1 μL of 10 μM universal forward primer, and 1 μL of 10 μM universal reverse primer. The universal forward primer has a 6-carboxyfluorescein fluorophore attached to the 5′ end that is used to detect the amplicon during peak height analysis. The qPCR cycling conditions were as follows: 37°C for 30 minutes, 95°C for 10 minutes, 60°C for 30 seconds, 95°C for 30 seconds, 67°C for 30 seconds (35 cycles), and 72°C for 20 minutes. Analysis of the MLPA PCR products for each gene was performed on an Applied Biosystems 3100-Avant genetic analyzer (Applied Biosystems) in a mixture of 10 μL of deionized formamide (Applied Biosystems), 1 μL of PCR product, and 0.5 μL of marker including a ROX-labeled internal size standard (ROX-500 GeneScan; Applied Biosystems) using POP-4 polymer (Applied Biosystems).
Table 1. Probe and Primer Target-Specific Sequences
Left Gene-Specific Sequence
Right Gene-Specific Sequence
After separation by capillary electrophoresis, peaks corresponding to each probe were identified by GeneMapper analysis (Applied Biosystems). Samples in which the smallest peak was < 100 relative fluorescent intensity units were not analyzed. The raw peak area for each probe in each sample was divided by the average raw peak area for all probes in that sample. This normalized peak area was then divided by the normalized peak area of the reference samples. Those control probes with a coefficient of variation > 20% were removed from the analysis. Only those samples with at least 3 control probes remaining were scored. Test probes > 1.25 were considered gained and those < 0.75 were considered to be lost. These levels were chosen empirically using the distribution of copy levels of control probes. Gain or loss for a chromosomal region was scored if at least 2 markers were gained or lost, respectively. Scoring was performed without knowledge of outcome, and without knowledge of 1p and 16q LOH status.
The 2 endpoints used in the current study were 8-year EFS and OS. EFS and OS curves were estimated using the Kaplan-Meier method and compared using the log-rank test. Relative risks (RRs) were calculated using the Cox proportional hazards model. Tests of correlation of 1q gain status and patient or disease characteristics were performed using the standard chi-square test for contingency tables.
A total of 212 patients met the patient and sample criteria for the current study; included were patients with NWTSG stage I (82 patients; 39%), stage II (54 patients; 25%), stage III (46 patients; 22%), and stage IV (30 patients; 14%) disease. This distribution is comparable to that noted overall in the NWTSG studies. The 8-year EFS estimate was 87% (95% confidence interval [95% CI], 82%-91%) for the entire cohort of 212 patients. Tumors from 58 of 212 patients (27%) displayed evidence of gain of 1q using the methods and criteria described above. The 8-year EFS rate was 76% (95% CI, 63%-85%) for those with 1q gain and 93% (95% CI, 87%-96%) for those who lacked 1q gain (P = .0024) (Fig. 1). The OS rate was 89% (95% CI, 78%-95%) for those with 1q gain and 98% (95% CI, 94%-99%) for those who lacked 1q gain (P = .0075) (Fig. 2). There were too few EFS events (total of 27) to analyze the effect of 1q gain within disease stage subsets. However, there was no indication that 1q gain correlated with disease stage because gain of 1q was identified in 20%, 31%, 37%, and 27% of patients with overall stages I, II, III, and IV disease, respectively (P = .16). Estimating the effect of 1q gain on EFS after stratification for stage, 1q gain was found to be associated with a significant increase in the risk of disease recurrence (risk ratio estimate, 2.72; P = .0089). For OS, tumors with 1q gain trended strongly toward an increased risk of death (risk ratio estimate, 3.08). However, this did not reach statistical significance (P = .067), possibly due to the small number of deaths overall (total of 9 deaths in the study cohort).
Similarly, the association with outcome was determined for 1p and 16q copy number loss. Retention of both 1p and 16q (no loss at either allele) was identified in 142 patients (estimated EFS rate, 89%; 95% CI, 83%-93%), isolated 16q loss was identified in 25 patients (estimated EFS, 84%; 95% CI, 63%-94%), isolated 1p loss was identified in 14 patients (estimated EFS, 79%; 95% CI, 47%-93%), and 7 patients had combined 1p and 16q copy number loss (estimated EFS, 71%; 95% CI, 26%-92% [P = .29]). There were too few cases with combined 1p and 16q loss to reliably assess whether this subset was associated with poorer EFS. It is interesting to note that in the current study, tumors with 1q gain were more likely to have 16q loss (42.6%) or 1p loss (32.5%) than tumors without 1q gain (7.9% and 5.4%, respectively; P < .0001 for both comparisons).
Copy number loss is not always equivalent to LOH, because LOH may not result in a net loss of genetic material. LOH resulting from somatic recombination is copy number neutral, as is LOH resulting from chromosome loss and reduplication of the remaining homologous chromosome. LOH data for 1p and/or 16q (performed using microsatellite analysis and reported previously) were available for many of the tumors analyzed in the current study. LOH data were available for 27% and 98%, respectively, of the samples tested in the current study for 1p and 16q copy number, and the concordance was 95% and 97%, respectively. The discordance was largely due to the identification of copy-neutral LOH in 2 of 7 cases with 1p LOH and 5 of 35 tumors with 16q LOH.
Approximately two-thirds of tumors with 1q gain were found to contain translocations between chromosomes 1 and another partner chromosome (Fig. 3A). Although several different chromosomal partners were noted in the cohort, the most common translocation partner was chromosome 16. The majority of the remaining patients with 1q gain contained the isochromosome 1q, i(1q), which results in gain of 1q and loss of 1p copy number (Fig. 3B). Similarly, approximately 40% of patients with 16q loss demonstrate t(1q;16q), all of which retained der(16) and lost der(1) (Fig. 3B). Many also contained duplication of the remaining intact chromosome 1, resulting in monosomy 16q, trisomy 1q, and 2 identical copies of 1p. This results in LOH for 1p (copy neutral) and 16q. It is interesting to note that in approximately one-third of tumors, 16q loss was a result of unbalanced translocations with other partner chromosomes. Such translocations involved a mixture of duplications and losses similar to that observed in t(1;16). Approximately 21% of tumors demonstrated loss of the entire chromosome 16.
The central goal of the COG Renal Tumor Committee is to increase the survival of patients with FHWT and minimize the toxicities associated with therapy. Although the clinical features of patient age and tumor stage are valuable predictors of the risk of disease recurrence in patients with FHWT, they have limited sensitivity and specificity. Novel biomarkers are needed to enhance the current risk stratification schema, thereby allowing the augmentation of therapy for patients at a high risk of disease recurrence and a reduction in therapy for patients with a low risk of disease recurrence. The current study sought to evaluate 1q gain as a possible biomarker for an unfavorable prognosis in patients with FHWT.
The association between chromosomal changes and disease recurrence in patients with FHWT has been suggested by several publications, all of which have demonstrated similar findings. In what to our knowledge is the largest study published to date, 127 of 195 patients who subsequently developed disease recurrence had abnormal karyotypes. The most common changes overall were gain of chromosomes 1q (28%), 8 (24%), and 12 (38%) and loss of 1p (13%), 11q (9%), 16q (19%), and 22 (10%). A stepwise Cox proportional hazards regression model demonstrated the significant independent predictors of risk to be 1q gain (RR, 3.4; P = .005), stage IV disease (RR, 5.0; P < .001), and monosomy 22 (RR, 5.9; P < .001). Copy number changes of 1p, 1q, and 16q were often found to be interrelated due to recurrent unbalanced chromosomal abnormalities. This group reported similar rates of unbalanced translocation with chromosome 16 and i1q formation, as was seen in the current study.
The NWTSG focused on LOH as a marker for disease recurrence. Analysis of 232 children registered on NWTS-3 and NWTS-4 whose tumors were FHWT or WT with anaplasia demonstrated LOH for markers on 1p and 16q in 12% and 17% of patients, respectively, and each was associated with a poorer recurrence-free survival (RFS) and OS when adjusted for stage of disease. This association was prospectively confirmed in 1727 informative FHWT cases registered on NWTS-5. An RR of disease recurrence of 1.56 and 1.49 was associated with 1p and 16q LOH, respectively, stratified by stage, with a similar RR noted for OS. When LOH for 1p and 16q were combined, the RR for death among the 970 patients with low-stage disease was 4.25, and that among the 686 patients with advanced disease (stage III or stage IV) was 2.66. On the basis of these studies, children registered on the current COG protocols were stratified according to their combined 1p and 16q LOH status. Patients with FHWT with combined LOH of 1p and 16q are treated on the current COG renal tumor protocols with intensified therapy, with the objective of improving the EFS and OS for these patients; those patients with both 1p and 16q LOH are treated more aggressively for each stage of disease. However, even if intensified therapy is shown to be beneficial for this group, this will not impact the majority of patients who will develop disease recurrence because 1p and 16q LOH are present in only 4.6% of FHWT, and are predictive of only 9.4% of all instances of disease recurrence. The RR of disease recurrence associated with LOH of 1p alone or 16q alone was not strong enough to warrant the risks of intensifying therapy in the current COG renal protocols.
In more recent years, classic comparative genomic hybridization (CGH) confirmed the most frequent alterations in patients with FHWT to be gain of 1q, 8, and 12 and loss of 1p, 11p, 16q, and 22. CGH analysis found only gain of 1q to be significantly associated with an adverse outcome. Gain of 1q was observed in 27 of 46 patients with recurrent FHWT (59%) compared with 5 of 21 patients with nonrecurrent FHWT (24%) who demonstrated a 1q gain (RR, 2.5; P = .019). Although the majority of tumors with 1q gain demonstrated gain of the entire long arm, 6 tumors demonstrated gain of smaller regions, with the smallest region of common gain spanning 1q21 to 1q25. In 8 cases (25%), 1p loss coexisted with 1q gain and, of 27 cases of disease recurrence with 1q gain, corresponding loss of either 1p or 16q was identified in 26% and 37% of cases, respectively. These observations support a strong association between 1q gain and poor outcome, and suggest that the unbalanced chromosomal abnormalities mentioned above result in 1q gain. These observations were confirmed by the analysis of 76 FHWT cases using bacterial artificial fhromosome (BAC)-based array CGH, which further narrowed the recurrent region of gain to 1q22 to 1q23.
A major limitation of the studies cited above was that none was performed in a prospectively identified patient cohort that was consistently treated on a cooperative group protocol. In the current study, the cohort was derived from a prospective clinical and biologic trial, and 1q gain was analyzed using MLPA, which is capable of measuring both gain and loss of potential biomarkers. MLPA is robust and flexible, can be reliably multiplexed, and is inexpensive. It does not require a source for comparison germline DNA and can be performed on archival tissue. MLPA has become rapidly accepted in the research community, and has entered the clinical realm. The current study included probes within the minimal region of gain (1q22-1q23) (Table 1), and was also able to include probes to 1p and 16q to correlate 1q gain with 1p and 16 loss.
The results of the current study indicate that gain of 1q was associated with a 17% absolute reduction (P = .0024) in the 8-year RFS and a 9% absolute reduction (P = .0075) in the 8-year OS. Furthermore, 1q gain was distributed relatively evenly among stage I to stage IV tumors. Although true within-stage analysis was not possible due to the small sample size for each disease stage, the prognostic significance of 1q gain remained (RR for disease recurrence, 2.72; P = .0089) when controlling for disease stage. The RR of death was slightly below usual significance levels (RR, 3.07; P = .067), but analysis was limited by the small number of deaths in the overall cohort (9 deaths of 212 subjects). Furthermore, the results of the current study confirm a strong association between 1q gain and 1p and 16q loss. The important gene or genes on 1q that contribute to the reduction in survival remains an enigma.
In conclusion, these findings suggest that 1q gain may indeed be useful for stratifying therapy in future therapeutic trials for patients with FHWT. This will require validation in an independent set of tumors and will be accomplished using the > 1400 patients registered on NWTS-5. This larger group of patients will enable the evaluation of smaller subsets based not only within disease stages, but also on 1p and 16q copy number and LOH status. Although the overall prognosis for patients with FHWT is good compared with that for many other malignant diseases, there is significant variation among patients with higher stages of disease. Approximately 15% of patients with stage III and 25% of patients with stage IV FHWT will experience disease recurrence when treated with current standard therapy. Given the significant heterogeneity of patients, especially within stage III and stage IV FHWT, the addition of 1q gain to the existing prognostic framework of clinical, pathologic, and biologic features has the potential to substantially improve the accuracy of risk stratification for appropriate therapy selection across all stages of FHWT. As a result, it may be possible to not only intensify therapy early for those patients at a higher risk of disease recurrence or death but also to decrease therapy for those patients with excellent RFS and OS who lack markers of high-risk disease. Given this significantly higher percentage of patients with 1q gain compared with LOH 1p and/or 16q, 1q gain has the potential to lead to significant improvements in patient outcomes if similar results are noted in validation studies.
Research was supported by grants from the National Institutes of Health to Dr. Perlman (CA155556A), to the National Wilms Tumor Study Group (CA-42326), the National Wilms Tumor Late Effects Study (CA-54498), and the Children's Oncology Group (CA-98543 and CA-98413).
CONFLICT OF INTEREST DISCLOSURES
Dr. Jennings is employed by the Ann and Robert H. Lurie Children's Hospital of Chicago. He has received payment for a plenary lecture from the Association for Molecular Pathology and was a panelist representing the College of American Pathologists at the National Comprehensive Cancer Network 2013 NCCN Policy Summit: Evolving Policy Issues in Oncology—Revisiting Biosimilars and Molecular Testing. Dr. Dome has received grant support from the National Institutes of Health (NIH) for Children's Oncology Group clinical trial leadership and has received royalties from St. Jude's Children's Research Hospital for anti-telomerase antibody 1. Dr. Grundy has received a grant from the NIH for the Children's Oncology Group Cooperative Group. Dr. Perlman has received an NIH grant as well as support from the NIH/National Cancer Institute for travel to meetings.