1 INTRODUCTION

According to the International Society of Paediatric Oncology (SIOP) staging criteria for renal tumours of childhood,¹ tumour rupture (TR) before surgery or intraoperatively results in upstaging to local stage III because of an increased risk of abdominal recurrence.² Local stage III requires intensification of postoperative treatment, including radiotherapy to flank or whole abdomen (WART).³ It has been shown that WART carries a high risk of long-term infertility and moderately increased lifelong risk for secondary tumours, particularly bowel cancer.⁴ Hence, the definition and diagnostic criteria for TR should be very accurate.

Intraoperative TR has clearly defined diagnostic criteria based on macroscopic appearances during surgery.⁵ Preoperative TR is more difficult to define,⁶ and there is no ‘gold standard’ for its diagnosis. Clinically, it may manifest with acute abdominal pain, haemorrhagic anaemia or shock,⁶ which may require an upfront surgery.⁷ However, none of these clinical symptoms can be used independently to predict TR because of insufficient positive predictive value, estimated between 16% and 54%.⁶ Moreover, it is also well known that TR may not be confirmed at pathological examination after neoadjuvant chemotherapy due to formation of a pseudocapsule, which may cover the site of rupture.^8-11

Radiological diagnostic criteria for preoperative TR are poorly defined, and in the absence of clinical signs, using radiological criteria alone to diagnose preoperative TR can be misleading.¹² Moreover, studies have shown that both diagnostic¹³ and preoperative¹⁴ computed tomography (CT) have low sensitivity and specificity in detecting preoperative TR regarding concomitant histological analysis. It is important to recognise the different approaches to renal tumour management between protocols of the Children's Oncology Group (COG) and the International Society of Paediatric Oncology Renal Tumour Study Group (SIOP-RTSG). Findings reported by COG mainly apply to the diagnostic scan, whereas the SIOP use of neoadjuvant chemotherapy provides two time points for assessment of preoperative scans.

The aim of this study is to describe the associations between radiological, surgical and pathology-detected TR in children with Wilms tumour (WT) in a nationwide multicentre clinical study.

2 PATIENTS AND METHODS

2.2 Inclusion and exclusion criteria

There were 712 patients registered in the IMPORT study from 20 centres between September 2012 and March 2021. Inclusion criteria were patients with newly diagnosed WT and indication of TR on the central radiology review (CRR) case report form (CRF) of the diagnostic cross-sectional abdominal imaging and/or on the surgical CRF. Exclusion criteria were patients with non-WT, registered at relapse, extrarenal site tumour and diagnosed abroad, or if the timing of TR was indicated on the surgical CRF as intraoperative (Figure 1). Clinical data for the included newly diagnosed patients with WT were extracted from the IMPORT study CRFs and central pathology review (CPR) letters. Correlation between radiology, surgery and pathology findings was made.

3 RESULTS

3.2 Tumour rupture on central radiology review

From the included cohort, 124/141 (86%) patients had images available for CRR. Imaging quality was recorded as 66 adequate, 22 suboptimal and 10 quality unrecorded. Ninety-eight patients had radiological signs suggestive of TR on the diagnostic cross-sectional abdominal imaging (63 MRI/35 CT). In 47/98 (48%), TR was limited to the renal fossa and in 51/98 (52%) involved peritoneum. Three patients had upfront surgery, 87/95 (92%) had findings of TR confirmed on the post-chemotherapy preoperative scan, whilst eight of 95 (8%) did not. Among 80/98 (82%) cases with TR on diagnostic CRR and available surgical forms, TR was confirmed on surgery in 29/80 (36%); on both surgery and pathology in 15/80 (15%), only on pathology in 13/80 (16%), not confirmed on either surgery or pathology (false positive) in 38/80 (48%) (Figure 3).

TR confirmed on pathology as a sole reason or in combination with other reasons (positive lymph nodes, incomplete tumour resection, tumour thrombus-related reasons) was responsible for stage III in 38/98 (39%) with subsequent flank or WART. Four of 98 (4%) were upstaged to stage III and received WART due to radiological TR confirmed at nephrectomy, but not on pathology. Twenty of 98 (20%) received flank or WART for stage III due to other reasons for stage III. Three of 98 (3%) with stage II received radiotherapy due to diffuse anaplasia on pathology (Table 2).

TABLE 2. Staging, radiotherapy and clinical outcomes for patients with tumour rupture on central radiology review.

	TR confirmed on surgery		TR not confirmed on surgery		Missing surgical forms		Total
	n	%	n	%	n	%	n	%
Number of patients	29	30	51	52	18	18	98	100
Finala abdominal stage
Stage I	0	0	17	33	2	11	19	19
Stage II	1	3	10	20	6	33	17	17
Stage III	28	97	24	47	10	56	62	63
Reasons for stage III
Histological features of TR with/without other reasonsb	20	71	12	50	6	60	38	61
No histological features of TR but upstaged only due to preoperative TR	4	14	0	0	0	0	4	6
Other histological reasons	4	14	12	50	4	40	20	32
Radiotherapy
Flank	8	26	14	27	4	22	26	27
WART	18	62	5	10	4	22	27	28
Missing postoperative treatment forms	3	10	6	12	2	11	11	11
Clinical outcomes
Distant relapse	2	7	3	6	1	6	6	6
Abdominal relapse	2	7	3	6	1	6	6	6
Combined relapse	0	0	1	2	0	0	1	1
Death	3	10	1	2	0	0	4	4

• Abbreviations: TR, tumour rupture; WART, whole abdominal radiotherapy.
• ^a Assigned by the oncology multidisciplinary team.
• ^b Positive lymph nodes, incomplete tumour resection, tumour thrombus-related reasons.

4 DISCUSSION

This study aimed to describe associations between TR detected on radiology, with subsequent findings at surgery and pathology and vice versa in a large cohort of children newly diagnosed with a renal tumour registered in a multicentre prospective clinical study that included CRR. We showed that diagnostic abdominal cross-sectional imaging has a high false-positive and false-negative rate for detection of TR 48% and 47%, respectively, with both surgery and pathology being the reference. These findings suggest that imaging should not be relied upon to detect TR nor used to make treatment decisions in this respect. It is important to understand the tumour anatomy, where it extends to, the presence of locoregional or distant disease and the treatment response, but imaging is not a reliable method for detection of TR using current radiological criteria. These are subjective and include perinephric or intraperitoneal fluid,^{6, 12, 13} intraperitoneal nodules,¹⁷ retroperitoneal tumour extension or ‘capsule breach’.¹³ Each of these features were considered by the reviewing radiologists as indicative of or suspicious for TR, but this needs further consensus and building work.

We suggest that the term ‘tumour rupture’ is a fundamentally misleading term to be used by radiologists. Instead, clearly defined terms such as ‘extracapsular tumour extension’ (meaning that tumour extends beyond the kidney and its own tumour pseudocapsule) and ‘intraperitoneal tumour spread’ (meaning that there are tumour deposits separate to the main tumour within the peritoneal cavity) should be adopted. ‘Radiological features highly suggestive of tumour rupture’ should only be used where a combination of findings is present: large volume ascites, extracapsular tumour extension and intraperitoneal tumour spread.

Our data show that there were nine of 141 (6%) children who had pathology stage I or II, but were upstaged to stage III and treated with radiotherapy due to preoperative TR. None of those were patients with peritoneal tumour implants. All these patients are in the first complete remission. Similar results have been shown in the study conducted by Le Rouzic et al.,¹⁴ where they have overtreated/stage V patients. Therefore, dilemma arises if this cohort of patients should be upstaged and treated with intensified chemotherapy and WART as stage III. Few studies have shown that the prognosis seems not to be better in patients treated with intensified postoperative chemotherapy and radiotherapy versus patients treated less aggressively for preoperative TR alone.^{6, 18} These studies are small series; therefore, larger clinical trials are required to determine whether preoperative TR suspected on imaging, clinical grounds or by the surgeon but without pathological confirmation has prognostic significance in the long-term survival for these patients.

Finally, our study showed that all six patients with upfront surgery due to acute bleeding had TR confirmed either on surgery or pathology or both, but only half of them had TR seen on radiology review. All these patients were assigned stage III and treated accordingly. The rate of the emergency surgery reported in literature is low, between 0.8%⁶ and 3%,⁷ and the predictive value of clinical signs of pain and haemorrhage seems to be low varying from 13%⁶ to 54%.¹⁹ We did not systematically collect data on clinical signs of TR at the time of diagnosis, and therefore cannot include this in our analysis.

A limitation of our study was that we did not perform an assessment of interobserver variation in diagnosing TR on CRR, and therefore are not able to establish specificity and sensitivity for CT and MRI in detecting TR. This is an important area for further research. Images were obtained from different centres, with different imaging protocols, on different scanners and in the case of MRI, on both 1.5 and 3 T, around 20% of the studies with deemed to be of suboptimal quality at CRR. Another limitation is that we had around 13% of patients with missing images for CRR and 13% missing surgical forms. Additionally, the documentation of preoperative TR on the surgical form might be biased if it was based on the radiological conclusions/MDT (multidisciplinary team) discussion regarding TR rather than intraoperative findings.

5 CONCLUSIONS

Radiology alone should not be used to define TR as it has not been shown to accurately correlate with surgery and pathology findings. Management and staging decisions should be based on multimodality approach with MDT considering clinical, radiological, surgical and pathology findings.

Future areas for research could be to identify accurate imaging markers that can detect TR, but this does rely on an accurate gold standard to validate against. Novel imaging biomarkers including multiparametric analysis and machine learning may be avenues for exploration, as the current imaging markers have proven insufficient.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.