Clinical Significance of Pulmonary Nodules Detected by CT and Not CXR in Patients Treated for Favorable Histology Wilms Tumor on National Wilms Tumor Studies-4 and 5: A Report from the Children’s Oncology Group

Abstract

Background

Metastatic lung disease in Wilms tumor (WT) patients was traditionally identified by chest radiograph (CXR). It is unclear whether patients with small lesions, detectable only by computed tomography (“CT-only” lesions), require the more intensive therapy, including doxorubicin and lung irradiation, given to patients with metastases detectable by CXR.

Procedures

This study involved 417 patients with favorable histology WT and isolated lung metastases (detected by CXR or CT) who were registered on National Wilms Tumor Study (NWTS)-4 or -5. Outcomes by method of detection (CXR vs. CT only), use of lung radiation, and 2- or 3-drug chemotherapy (dactinomycin and vincristine +/− doxorubicin) were determined and compared using the log-rank test.

Results

There were 231 patients with lung lesions detected by CXR and 186 by CT only. Of the patients with CT-only nodules, 37 received only 2 drugs and 101 did not receive lung radiation. Five-year event-free survival (EFS) was greater for patients receiving 3 drugs (including doxorubicin) with or without lung radiation than for those receiving 2 drugs (80% vs. 56%; p=0.004). There was no difference seen in 5-year overall survival (OS) between the 3-drug and 2-drug subsets (87% vs 86%; p=0.91). There were no significant differences in EFS (82% vs. 72%; p=0.13) or OS (91% vs. 83%; p = 0.46) for patients with CT-only nodules whether they received lung radiation or not.

Conclusions

Our results suggest that patients with CT-only lung lesions may have improved EFS but not OS from the addition of doxorubicin but do not appear to benefit from pulmonary radiation.

Introduction

Stage IV Wilms tumor is defined by the presence of hematogenous metastases to the lung, liver, bone, brain, or other site, with the lung being the most common site of metastasis. The protocol for the first National Wilms Tumor Study (NWTS), which began in 1969, used the chest radiograph (CXR) to detect pulmonary metastases. The introduction of computed tomography (CT) has necessitated a reconsideration of this definition of metastatic disease, as many patients have been found to have lung lesions detectable on CT scan which were not detectable on CXR (1–5). CT scans have higher sensitivity than CXR and are able to detect lesions smaller than 1 cm that are too small to be seen on X-ray, but they also carry the disadvantages of high interreader variability (6;7) and questionable specificity for metastatic lesions, particularly the smallest ones (3,4,7).

Controversy developed regarding the treatment of patients with CT-only lesions because the probability of a CT-detected lesion being metastatic disease was unknown. Lesions too small to be detected on CXR may not require additional intensive therapy, thus avoiding the consequent acute and late toxicities. Complications associated with doxorubicin include arrhythmias, chronic or acute heart failure (especially with higher doses), and second malignancies. Complications associated with pulmonary irradiation (RT) include radiation pneumonitis and interstitial lung disease, scoliosis, second malignancies, and exacerbation of the cardiotoxic effect of doxorubicin.

Two previous studies investigating the role of more intensive treatment in patients with CT-only nodules enrolled in NWTS-3 and -4 did not show any significant differences in relapse-free survival (RFS) or overall survival (OS) between patients with CT-only nodules treated according to local tumor stage and those treated as Stage IV (5;8). Both studies had small sample sizes (27 Patients with CT-only nodules with favorable histology in the first study and 90 in the second). We undertook the present study to clarify these findings by using data from a larger number of patients (186 Patients with CT-only nodules) with favorable histology Wilms tumor enrolled in NWTS-4 and -5 who had metastases to the lungs only, in which treatment of patients with CT-only nodules as Stage IV or according to local tumor stage was at the discretion of the investigator (9–11).

Methods

Patients

This study involved 417 patients from the multi-institutional randomized trials National Wilms Tumor Study (NWTS)-4, conducted between August 1986 and August 1995, (165 patients) and NWTS-5, conducted between August 1995 and May 2002 (252 patients). The protocol details and study outcomes of NWTS-4 and 5 have been reported (9–11). For both studies, the protocol mandated a two-view chest x-ray as part of staging studies. Many patients also underwent CT scanning as per institutional practice but this was not required by the protocol. For both studies, CXRs and CT scans were reported by the local radiologist and were not further reviewed. The basic treatment guidelines for Stage I and II favorable histology tumors included nephrectomy followed by 2-drug chemotherapy (vincristine and dactinomycin). Stage III patients underwent nephrectomy, treatment with 3-drug chemotherapy (vincristine, dactinomycin, and doxorubicin), and abdominal RT post-nephrectomy. Stage IV patients with pulmonary metastases underwent nephrectomy, treatment with 3-drug chemotherapy, abdominal RT (if local tumor was Stage III) and 1200 cGy bilateral pulmonary RT. For patients with CT-only lung lesions, treatment was according to the local tumor stage (I, II, or III) and the addition of lung radiotherapy and/or doxorubicin (if local stage I or II) was at the discretion of the investigator. All patients on NWTS-4 and -5, or their parents or legal guardians, gave informed consent. Institutional IRB approval was required for participation in the studies.

Patients were included in this analysis if they (1) were enrolled in NWTS-4 or -5; (2) had been diagnosed to have favorable histology Wilms tumor; (3) had a radiology checklist on file; (4) had been diagnosed to have lung metastases by two-view CXR or CT scan; and (5) had no metastases to other sites although intra abdominal lymph node involvement was not considered metastatic. This study included those NWTS-4 patients analyzed in the 1999 study by Meisel et al. (8), which analyzed patients from NWTS-3 and -4. NWTS-3 patients were not included in the present study because the treatment regimens used in NWTS-3, as well as the quality of the CT scans used at the time, were not comparable to those in NWTS-4 and -5. In NWTS-4, patients were randomized to four treatment dosing schedules utilizing the same drugs, among which no significant survival differences were found (9;10). The treatments in NWTS-5 used the same drug combinations and were based on the most cost-effective dosing schedule in NWTS-4. Thus, patients from the two groups were combined for this analysis.

Statistical Methods

Event-free survival (EFS) was defined as the time from study entry to the first occurrence of progression, relapse after response or death from any cause as a first event. Overall survival (OS) was defined as the time from study entry to death from any cause. Follow-up for patients not experiencing an event was censored at the last patient contact. Estimates of event-free and overall survival were determined using the Kaplan-Meier method (12). Comparisons of EFS and OS distributions among patient subgroups were based on the log-rank test (13).

Results

Of 2555 patients with favorable histology Wilms tumor enrolled in NWTS-4, 729 (29%) received both CXR and chest CT scan. For 1621 patients, radiology checklists were unavailable regarding whether they had received CXR, CT, or both. Of 2331 patients with favorable histology Wilms tumor enrolled in NWTS-5, 1763 (76%) received both CXR and chest CT scan. For 90 patients, radiology checklists did not indicate which modalities were used. There were 417 patients on NWTS-4 and -5 who met the inclusion criteria. Of these, a total of 231 patients had lung lesions detectable on CXR and 186 had CT-only lesions. Among the 165 patients from NWTS-4, 108 were positive for lung lesions on CXR and 57 (35%) were positive on CT scan only (‘CT-only’ patients). Among the 252 patients from NWTS-5, 123 were positive on CXR and 129 (51%) were positive on CT scan only (Table I). Outcomes were similar between NWTS-4 and -5 patients, so data were combined for most analyses.

Table I.

Distribution of patients by study and diagnostic modality.

Lung metastasis diagnosis	Study status*			Total
	Initial Nephrectomy	Pre-nephrectomy chemotherapy	Bilateral Wilms’ tumor
NWTS-4 (N = 165)
X-ray(+)	68	36	4	108
CT(+), X-ray(−)	51	5	1	57
NWTS-5 (N = 252)
X-ray(+)	82	37	4	123
CT(+), X-ray(−)	109	17	3	129
Total	310	95	12	417

^*Bilateral patients may have received pre-operative treatment

All patients with positive CXR had positive CT scans except eight patients who did not receive a CT exam, as stated on the radiology checklist. All patients designated ‘CT-only’ had a positive CT-scan and a negative CXR. Eighty-five patients who had received 6 weeks of pre-nephrectomy chemotherapy, as well as 12 patients with bilateral Wilms tumor were also included in the initial analyses (Table I).

Patients with CT-only nodules were younger on average at diagnosis than CXR-positive patients (54.2 months vs. 61.6 months; p=0.02). Tumor weight and the percentage of male patients did not differ significantly between CXR-positive and Patients with CT-only nodules. Data on the size and number of lung lesions, based on CT scan, were available for about half of all CXR-positive patients and Patients with CT-only nodules. The median size of the largest lesion was 6 mm for Patients with CT-only nodules and 16.5 mm for CXR-positive patients.

For patients with CT-only lung lesions, treatment specified by the NWTS protocol was determined by local tumor stage (I, II, or III), with the addition of pulmonary RT and/or doxorubicin (if local stage I or II) at the discretion of the institutional investigator. Thus, about half of patients with CT-only nodules were designated and treated as Stage IV (56% in NWTS-4 and 42% in NWTS-5), while the remainder were designated and treated as Stage I, II, or III (see Table II). Five percent of CXR-positive patients were treated with 2 drugs (vincristine and dactinomycin) compared to 21% of patients with CT-only nodules. The remaining 79% of the patients with CT-only nodules were treated with 3 drugs (vincristine, dactinomycin and doxorubicin). Similarly, 91% of CXR-positive patients were given lung radiotherapy, whereas only 41% of patients with CT-only nodules received lung radiotherapy.

Table II.

Lung metastasis diagnosis by stage and study.

Lung metastasis diagnosis	Stage					Total
	I	II	III	IV	V
NWTS-4 (N = 165)
X-ray(+)	0	2	0	102	4	108
CT(+), X-ray(−)	4	7	13	32	1	57
NWTS-5 (N = 252)
X-ray(+)	1	1	3	113	5	123
CT(+), X-ray(−)	5	22	45	54	3	129
Total	10	32	61	301	13	417

EFS and OS, of patients with CT-only pulmonary lesions regardless of treatment, were not significantly different for patients treated with initial nephrectomy versus pre-operative chemotherapy, so outcomes for these groups were analyzed together. It was elected to exclude the four patients with CT-only nodules with bilateral tumors from the outcome analyses because issues arising from multiple local tumors were considered a confounding factor when assessing outcome from the pulmonary lesions.

Five-year EFS was significantly higher for patients with CT-only nodules who received 3-drug chemotherapy with or without lung radiation than for patients who received 2-drug chemotherapy with or without lung radiation (79.7% vs. 56.0%; p=0.0039) (Table 3). When adjusted for use of lung RT, the difference remained (p=0.03). There were no differences in OS whether treated with 2 or 3 drugs (p=0.91, Table III).

Table III.

Effect of chemotherapy on outcomes for patients diagnosed to have lung metastasis by CT only.

Chemotherapy	# pts	Event-free survival % at (95% CI)		p-value	Overall survival % at(95% CI)		p-value
		2 yrs	5 yrs		2 yrs	5 yrs
2 drugs	37	59.8 (41.6,73.9)	56.0 (37.7,70.9)		91.3 (75.5,97.1)	86.0 (65.2,94.8)
3 drugs	145	84.2 (77.0,89.3)	79.7 (71.3,85.9)	0.0039	94.0 (88.2,97.0)	87.0 (78.6,92.3)	0.91

There was a non-significant difference in 5-yr EFS for patients treated with or without lung radiation (81.0% vs. 70.1%; p=0.11) (Table IV), but this difference disappeared when the analysis was adjusted for the chemotherapy regimen delivered (p=0.52). Similarly, there was no difference in OS whether lung RT was used (p=0.73).

Table IV.

Effect of lung irradiation on outcome for patients diagnosed to have lung metastasis by CT only.

Lung RT	# pts	Event-free survival% at (95% CI)		p-value	Overall survival% at (95% CI)		p-value
		2 yrs	5 yrs		2 yrs	5 yrs
No	105	75.0(65.3,82.4)	70.1(59.3,78.6)		94.3(86.8,97.6)	83.7(71.8,90.9)
Yes	77	84.8(74.2,91.3)	81.0(69.3,88.6)	0.11	91.9(82.9,96.3)	90.0(80.1,95.2)	0.73

Patterns of recurrence were examined for the 2- versus 3-drug treatment groups. Eleven of 14 (79%) recurrences in the 2-drug group were in the lung versus three in the abdomen. There was one toxic death. In comparison, 12 of 20 (60%) in the 3-drug group relapsed in the lung versus eight in the abdomen, while there were four toxic deaths and three with progressive disease. These numbers were too few to allow meaningful statistical analysis. Causes of death were also contrasted. Of four deaths in the 2-drug group, two (50%) were due to tumor, one was due to toxicity and one was unknown. In the 3-drug group, 11 of 15 (73%) deaths were attributed to tumor, two to toxicity and two to other causes.

Discussion

This analysis demonstrates that the 5-yr EFS percentage of patients with CT-only nodules with favorable histology Wilms tumor without other sites of metastasis may have been improved by the addition of doxorubicin in addition to vincristine and dactinomycin. However, the addition of bilateral lung radiation to the 3-drug chemotherapy regimen did not significantly affect 5-yr EFS. In addition, the OS of patients did not differ by chemotherapy regimen or lung RT.

There is some confounding of the outcomes when examining the effect of each factor because almost all patients who received only two drugs did not get RT but about half of those treated with three drugs did get RT but not in a randomized fashion. Although two previous studies investigating the role of more intensive treatment in patients with CT-only nodules enrolled in NWTS-3 and -4 did not show any significant differences in relapse-free survival or overall survival between patients treated according to local tumor stage and those treated as Stage IV (5;8), the present study includes more than twice the number of patients as the previous studies. We believe that the difference in EFS demonstrated in the current study is real for several reasons. First, although only 39 patients in our study who were included in the CT-only group received 2-drug therapy, the EFS percentage found here (56%) is comparable to a previously reported relapse-free survival percentage, 42%, for patients with pulmonary lesions detected by CXR treated with 2 drugs (and pulmonary RT) on NWTS-2 before the advent of doxorubicin (14). This comparable result suggests that the poor outcome of this group is real and increases the confidence in the accuracy of the outcomes for the current group of only 39 patients. Secondly, the benefit of using more intensive chemotherapy for patients with CT-only nodules was also suggested by results of the second United Kingdom Children’s Cancer Study Group Wilms tumor study (UKW2), which found that Stage I patients, who were normally treated with vincristine alone, had a higher relapse rate when CT-only lesions were present (15). Thirdly, the predominance of lung recurrences (79%) in the group treated with only two drugs is consistent with under-treatment of the CT-only lung lesions.

While the addition of doxorubicin was associated with improved EFS in this report, OS was not significantly affected. Given the high rate of OS it is possible that the sample size was insufficient to permit detection of small increments in survival. Alternatively, 5-yr OS might not be impacted due to the ability to salvage patients with recurrent disease. However, results of both low- and high-risk relapse studies from NWTS-5 suggest that only 70% of low-risk (16) and 42% of high-risk (17) patients can be salvaged, implying that an improved event-free survival should translate into an improved OS. It is also possible that an increased survival rate brought about by treatment with 3 drugs is balanced by an increase in toxic deaths due to doxorubicin and/or pulmonary RT, but the data do not support this possibility with the majority of deaths in both groups being due to tumor. Our data do not demonstrate a benefit to patients with CT-only nodules from the use of bilateral lung radiotherapy. Although there was a trend to improved EFS percentage in the patients who received lung RT, this trend completely disappeared when results were adjusted for chemotherapy regimen.

These results are in contrast to those reported by Nicolin et al based on the UKW2 and UKW3 trials which showed in a non-randomized fashion that those stage IV patients treated without pulmonary irradiation had a worse outcome despite the fact that all were treated with three drugs including doxorubicin at 300 mg/m² (18). The greatest distinction between this study from the United Kingdom and ours is that the UK study included children with pulmonary metastases detected by chest x-ray in contrast to our study which only included patients with pulmonary lesions small enough to be detected only on CT scan. Our interpretation of these combined studies is that patients with pulmonary metastases of any size benefit from the addition of doxorubicin to two-drug therapy, but that pulmonary irradiation may only be beneficial for larger lesions. Unfortunately neither of these retrospective studies can define the limits of these categories.

An additional difference between the NWTS and UKWT studies, although not one which would affect the conclusions herein, is in the dose of doxorubicin utilized. In the UKWT studies this has consistently been 300 mg/m2 for patients with metastatic disease (18). In the NWTS-4 study, patients were randomized to four regimens which effectively resulted in the use of 150 mg/m2 for half of the patients and 250 mg/m2 for the remainder. The benefit of doxorubicin for patients with smaller CT-only lesions was not obscured by the use of smaller doses in our study, but in the future, it may be possible to look at dose response using patients from the NWTS and the SIOP.

CT scans have higher sensitivity than chest radiographs and are able to detect lesions smaller than 1 cm that are too small to be seen on CXR, but they also carry the disadvantages of high inter-reader variability (6;7) and questionable specificity for metastatic lesions, particularly the smallest ones. CT-detected lung lesions that look like metastases may instead be infectious granulomata, normal intrapulmonary lymph nodes, hamartomas, round pneumonia, inflammatory pseudotumor, or round atelectasis (19;20). The latter is especially common if the CT is done under general anesthetic (21).

Studies to determine the percentage of CT lung lesions that actually represent metastatic disease have yielded conflicting results. Some studies assert that a significant percentage of lesions are metastatic tumor (22;23), while others assert that most lesions are benign, even in children with known cancer (3;4;7;19;20). Of the 42 patients with CT-only nodules who had undergone lung lesion biopsy at the attending physician’s discretion in NWTS-5, 82% of patients with isolated lesions and 69% of patients with multiple lesions had histologically-proven lung metastases (24). In a study of 24 patients with bone or soft tissue sarcoma who underwent biopsy of lung lesions, 41.7% were positive for metastatic disease (25).

This study did not identify specific criteria that can be used to determine whether a given patient’s lung lesions represented metastatic disease. A more specific classification of patients with lung lesions might be possible using largest lesion diameter and/or sum of lesion diameters, although some recent studies have found that larger lesions are not more likely to be metastatic in children with cancer (7;20;25). The current Children’s Oncology Group (COG) Renal Tumor Biology study seeks to generate consistent data to determine if there is a correlation between lesion size and the survival benefits of different treatments.

Of the patients in this study with pulmonary metastases who were studied by both imaging modalities, there was a much larger percentage of patients with CT-only nodules in NWTS-5 relative to NWTS-4. This may have been due in part to the increased awareness of the potential treatment significance of CT-only lung lesions (5), or the use of lower thickness image slices over time, although the major factor was probably the increased use of CT imaging in general.

The results of this retrospective study suggest that CT-only lung lesions have clinical significance in patients with favorable histology Wilms tumor. These data suggest that CT-only lung lesions should not be disregarded in patients who would otherwise receive 2-drug therapy, because the addition of doxorubicin to chemotherapy may reduce the risk for recurrence. As this was a small and retrospective study, the results must be interpreted with caution, and particularly for those already treated with three drugs, the avoidance of radiation must be very carefully considered.

Pertinant to our findings is the recent report by Smets et al concerning patients with CT-only lung nodules on the SIOP 2001 Wilms tumor trial (26). In contrast to our findings, they found no difference in outcome between 37 of 103 CT-only Wilms tumor patients treated initially as localized tumor stages compared with 66 similar patients who received initial treatment for metastatic disease. However, the majority of children treated as local disease received doxorubicin postoperatively, as did those treated as metastatic disease, so it is not possible to separate out the effects of doxorubicin or radiation. Furthermore, the outcomes for both groups of CT-only patients were significantly inferior to those with true unilateral localized disease and was not significantly different than those with true metastatic disease, further supporting that chest CT scan adds prognostic information in newly diagnosed patients with Wilms tumor.

Our data are perhaps best used to inform further prospective clinical trials. Current prospective studies by the Children’s Oncology Group may refine the treatment of patients with CT-only nodules by more exactly determining the appropriate use of pulmonary RT and by further optimizing chemotherapy drug regimens. The specific aim of the current trial for metastatic Wilms tumor is to test the safety of omission of pulmonary radiation for rapid early responders, those whose pulmonary lesions disappear by week six of therapy.