Management of malignant ureteric obstruction with ureteric stenting or percutaneous nephrostomy


 Malignant ureteric obstruction (MUO) is frequently encountered in patients with advanced cancers. In the largest study to date, assessing 852 patients from across Scotland, the authors demonstrated the presence of MUO as a marker of advanced disease across cancer types, with poor survival for many patients, even with intervention. There is uncertainty in optimal management of this condition, with marked differences in management between hospitals. Treatment to relieve the obstruction does not guarantee either improvement in kidney function or progression to further oncological treatment. The authors have developed a prognostic tool to estimate outcomes after intervention for MUO, and advocate its use for clinicians along with other data presented for patient counselling.



Introduction
Malignant ureteric obstruction (MUO) is intrinsic or extrinsic obstruction of one or both ureters by malignancy.The incidence of MUO is poorly captured; however, has been reported in up to 4% of patients with advanced malignancy [1][2][3] .Unlike other oncological emergencies, such as malignant spinal cord compression, there is no standard approach to its management.Intervention to decompress the kidney(s) via percutaneous nephrostomy (PCN) or ureteric stenting (US) may be undertaken to facilitate treatment of urosepsis or significant electrolyte disturbance, for palliative treatment of intractable pain, or in an attempt to improve renal function and facilitate oncological treatment.These interventions, however, may be associated with morbidity, mortality, and have quality-of-life implications [4][5][6][7][8][9][10][11][12][13][14] .
Given the potential negative impact and unpredictable overall survival (OS) after intervention, it is important to determine which patients benefit most.Several small retrospective studies [15][16][17][18][19] have demonstrated that survival in this cohort is poor.One study 15 of 53 patients demonstrated that the primary cancer site, laterality, creatinine concentration at presentation, and whether the patient went on to receive oncological treatment predicted outcome after US for MUO (PLaCT score).Although this was subsequently validated in a 300-patient case series across 8 centres, critically this did not include patients managed with PCN, and the separation of prognosis into good, intermediate, and poor groups in that study is unlikely to provide the granularity of analysis needed to allow adequate patient counselling 20 .
The authors sought to define patient outcomes after US or PCN for MUO in a large multicentre national cohort, and to assess patient and clinical factors with respect to OS.A further aim was to develop and validate a scoring system using parameters available at the time of presentation to identify patients with short OS (90 days or less) in whom relief of obstruction may not be indicated.

Methods
As an audit of standard of care, the study did not require specific ethical approval, in accordance with National Health Service Health Research Authority guidance.Caldicott Guardian approval was sought before local data collection.The study is reported in line with STROBE guidelines.
Consecutive patients with locally advanced or metastatic cancer undergoing primary US or PCN insertion for MUO from six health boards across Scotland between 2008 and 2020 were included in this study.Patients were identified locally as having undergone US or PCN insertion from prospectively recorded operation codes (supplementary material).Relevant clinicopathological data were extracted to a standardized pro forma (supplementary material).Data exclusions are shown in Fig. 1.
Primary outcomes were OS after decompression, and renal function at 3 months, specifically the proportion of the cohort with an estimated glomerular filtration rate (eGFR) over 45 ml per min per 1.73 m 2 , in accordance with the Kidney Disease Improving Global Outcomesk (KDIGO) class of initial renal failure 21 .This eGFR threshold was chosen as one that may allow chemotherapy to be considered.Secondary outcomes were return to oncological treatment (surgery, systemic therapy or radiotherapy, regardless of intent), and the development of a scoring system (Scottish MUO Score) to predict patients with a survival time of 90 days or less, and model its performance in prediction of overall survival at 30 days as well as 6 and 12 months.The performance of the Scottish MUO Score was compared with that of the PLaCT score 15,20 , which is used specifically in the setting of MUO, the modified Glasgow Prognostic Score (mGPS) 22,23 , and the C-reactive protein to albumin ratio (CRP : Alb) [24][25][26][27][28][29][30][31] .
Disease stage was broadly categorized as either locally advanced or metastatic, in accordance with the AJCC 8th edition for each cancer type.Locally advanced disease was defined as tumours of urological, gynaecological or gastrointestinal origin, radiologically staged as at least T2, causing intrinsic or extrinsic MUO owing to the mass effect of the primary tumour.The rationale for intervention was determined from contemporaneous medical records.

Statistical analysis
All statistical analysis was performed using R version 4.2.0 (R Foundation for Statistical Computing, Vienna, Austria) and RStudio version 2022.02.3 (Posit, Boston, MA, USA).P < 0.050 was considered significant.Details of handling of missing variables, imputation, and model development are provided in the supplementary material.

Results
PCN or US was performed for MUO in 852 patients during the study interval, with a median follow-up of 7 (i.q.r.2-20) months.Demographic and clinical data are shown in Table 1.The cancers were of urological origin in 466 patients (54.7%), gynaecological in 192 (22.5%), and gastrointestinal in 124 (14.6%).The nature of the disease causing ureteric obstruction is summarized in Fig. 2. Intervention was undertaken in patients 124 (14.6%) even though they had an eGFR exceeding 60 ml per min per 1.73 m 2 , with marked differences in the rates of initial US or PCN between centres (Fig. S1).Data were available regarding the primary rationale for intervention in 584 of the cohort (Table 1).
At 3 months after intervention, 378 patients (44.4%) had over 20% improvement in creatinine level, whereas 227 (26.6%) had no improvement.Renal function at 3 months stratified by class of initial renal failure is shown in Fig. 3.The proportion of individuals with an eGFR of at least 45 ml per min per 1.73 m 2 at 3 months varied significantly according to initial class of renal failure (P < 0.001), with a significant proportion of patients with an eGFR in the range of chronic kidney disease (CKD) 5, CKD4, and CKD3b showing an improvement in eGFR to 45 ml per min per 1.73 m 2 or higher (30.1, 35.3, and 51.6% respectively).Of patients with an eGFR in the normal or CKD3a range before intervention, 9.6% had a reduction in eGFR to below 45 ml per min per 1.73 m 2 .
Patient, radiological, haematological, and biochemical parameters relevant to OS were assessed by Cox regression (Table 2).In the adjusted model, age, metastatic disease at time of intervention, bilateral hydronephrosis, low haemoglobin, high potassium, high white cell count, and high CRP level were associated with worse OS.Although there were differences in OS based on cancer type on unadjusted analysis, after adjustment this only held true for prostate cancer and lymphoma.

Development of the Scottish MUO Score
Demonstration that cancer type is not the primary driver of survival provides evidence supporting the use of pan-cancer scores in patients with MUO.The presence and distribution of missing variables were assessed, and were then imputed, as described in the supplementary material and shown in Fig. S2.
After development, comparison of clinical and biochemical parameters between discovery and validation cohorts is reported in Table S1.
Stepwise regression identified that a model including age, CRP, haemoglobin, creatinine, potassium, sodium, cancer type, metastatic disease, and bilateral hydronephrosis was associated with the lowest information loss in determining OS in the discovery cohort.K-fold cross-validation was then used to determine the weightings of each of these variables in a predictive model for 90-day survival (supplementary material).This model is simplified for general use at: https://webapps.igc.ed.ac.uk/world/research/muo_calculator/.As well as calculating the Scottish MUO Score, the website also provides prediction of 30-day, 90-day, 6-month, and 12-month OS based on results from the discovery cohort (supplementary material).

Model performance
Model performance was assessed in discovery and validation cohorts by means of receiver operating characteristic (ROC) curves for the primary outcome of 90-day OS (Fig. 4 and Table 3), along with secondary outcomes of 6-month (Table S2a), 1-year (Table S3a), and 30-day OS (Table S4a).
To explore differences in specificity and sensitivity between scores, OS was compared in the discovery cohort for each of the scoring systems (Tables 4, S2b, 3b, and 4b).A range of OS outcomes was represented across deciles of the Scottish MUO Score (1.7-73.3%died within 90 days, and 15.0-85.0%died within 6 months), whereas CRP : Alb, PLaCT score, and mGPS were less able to personalize OS assessment as they had a narrower range of outcome predictions.

Discussion
This study analysed the largest detailed cohort of patients undergoing intervention for MUO, demonstrating across a multicentre national cohort that there is variability in the use of PCN and US.These differences highlight the uncertainty clinicians face regarding if, when, and how to intervene in MUO.It is clear, however, that MUO is associated with poor prognosis, with 89% of patients dying during follow-up and the majority having no improvement in renal function or receiving further oncological treatment.
Although prognosis is poor overall (including 8.5% of patients surviving less than 30 days), there is also a large group (38%) surviving for over 1 year after intervention.To differentiate these groups, clinical factors associated with worse OS were identified.The presence of metastatic disease had a negative impact on survival.Although it was not possible to assess such factors in the present study, disease stage and nature of obstruction are likely to affect treatment options, which will be the focus of future work.Nonetheless, the presence or absence of metastatic disease is a key stratifier in the MUO Score.In adjusted Cox regression analysis, only prostate cancer and lymphoma were associated with OS independently of the presence or absence of metastatic disease.This suggests that MUO may be a marker of advanced disease independent of the underlying tumour biology of different cancer types, supporting the use of a pan-cancer survival score in patient counselling before intervention.
The Scottish MUO Score developed and validated in this study includes parameters readily accessible in routine clinical practice, and the online platform allows straightforward calculation.It is also noted that, to a clinician and patient, life expectancy may be of greater relevance.Uniquely, the online platform uses routine clinical parameters to provide an assessment of 30-day, 90-day, 6-month, and 12-month OS from the discovery cohort, providing a personalized assessment of survival to guide patient counselling and decision-making.
The Scottish MUO Score compares favourably with other prognostic scores 15,22,23 .The attractiveness of the other scores is their simplicity, but they may fail to adequately discriminate between differences in outcomes 15,20 .To a patient contemplating intervention, the ability to determine whether they are going to be alive at defined time points is of paramount importance, and the Scottish MUO Score allows this.

Fig. 2 Nature of lesion causing obstruction for each cancer type, and whether the disease was locally advanced or metastatic at the time of intervention
Metastatic disease is subdivided according to the nature of the lesion causing ureteric obstruction: local disease (including adjacent nodal stations contiguously involved with primary tumour), regional lymph nodes not in continuity with the primary tumour, or non-regional lymph nodes/distant metastases.classified a high proportion of individuals in the poor prognostic groups.Although this leads to high sensitivity, patients may outlive this predicted life expectancy and might have been counselled against intervention.The lack of resolution about an individual's prognosis makes these scores unhelpful in a real-world clinical setting.The Scottish MUO Score discriminated patients with better OS; many patients with a low CRP : Alb died within 90 days (19.2% discovery cohort and 12.7% validation cohort CRP : Alb below 2), but this occurred only in 2 of 91 patients (2.2%) in the combined cohorts with a Scottish MUO Score of less than −3.0.

Dead
Decision-making in this area is complex, and a rationale for intervention for palliation remains in certain patients even when survival is expected to be poor.Potential immediately life-threatening indications, including sepsis and AKI, were present in 9.4 and 57.6% of the patients respectively, where intervention may be justified in the face of imminent death, even with the prospect of poor oncological outcomes.Indeed, although most patients did not have a significant improvement in renal function, there was a sizeable proportion of patients in this cohort with an arguably life-threatening eGFR of less than 15 ml per min per 1.73 m 2 who showed an improvement to over 45 ml per min per 1.73 m 2 .an understanding of the survival benefit with intervention is important.Understanding potential benefits allows patient-centred discussion of the balance of risks and benefits, particularly in view of the morbidity associated with intervention [3][4][5][6][7][8][9][10][11][12][13] .Even among teams unfamiliar with the individual patient, the Scottish MUO Score can aid this discussion.Although the proportion of patients in this cohort who received further oncological treatment is similar to that in previous reports 32 , the data do not allow assessment of quality-of-life or outcomes without intervention, which is the subject of an ongoing prospective study.
Although the inherent bias of case (and procedure) selection between health boards is likely to at account least partly for the reported differences in survival, the heterogeneous approach to management captured in this multicentre study is a particular strength.As a retrospective review, however, it was not possible to ascertain whether oncological treatment was completed, or to distinguish whether treatment was undertaken with curative or palliative intent.Missing from this cohort (and in other studies to    date) are the patients who do not go on to receive intervention, and the rationale for this decision.The present study is much larger than previous studies of patients with MUO, but it remains underpowered to subclassify each cancer type by stage or grade to fully appreciate their effect on OS.Further work to look at cancer-specific outcomes is warranted, particularly to determine whether bespoke scores for each cancer type could give additional prognostic information beyond the proposed pan-cancer Scottish MUO Score.Furthermore, differences in stage classification between cancers and their associated treatment options prevented their inclusion in the model, although this information could potentially be included in larger individual-cancer cohorts.The score also requires adequately powered external validation, particularly given that imputation was used for missing variables.Although definitions of curative or palliative treatment vary between specialties and across cancer types, further subgroup analyses would be of interest, and are the subject of an ongoing prospective study that also examines ethnicity, co-morbidity, readmission rates, rationale for the modality of decompression chosen, and quality of life.MUO is a marker of advanced disease and survival in these patients is poor.Differences in approach to management and outcomes across Scottish institutions highlight the uncertainty in optimal management of this condition.Although there will always be a specific rationale for intervention in patients with MUO, improvement in renal function or progression to additional oncological treatments is not guaranteed.Understanding survival after intervention for MUO is key to decision-making regarding whether to perform US or PCN, and the Scottish MUO Score (available at https://webapps.igc.ed.ac.uk/world/research/muo_ calculator/) can predict survival outcomes, aid patient counselling, and help deliver a patient-centred approach to treatment.

Fig. 3
Fig. 3 Relationship between outcome at 3 months and KDIGO class of renal failure before interventionKidney Disease Improving Global Outcomes (KDIGO) class of renal failure was determined from the estimated glomerular filtration rate (eGFR).An eGFR of 45 ml per min per 1.73 m 2 was chosen as the threshold that might allow future chemotherapy.

Fig. 4
Fig. 4 Receiver operating characteristic (ROC) curves for 3-month survival (90 days or less) according to scoring system a Discovery and b validation cohort.MUO, Malignant ureteric obstruction; mGPS, modified Glasgow Prognostic Score; CRP : Alb, C-reactive protein to albumin ratio.

Table 1 (
continued) The PLaCT score and mGPS

Table 3 Comparison of model performance for 3-month survival (90 days or less) in discovery and validation cohorts
Values in parentheses are standard errors.AUC, area under the curve; MUO, malignant ureteric obstruction; NPV, negative predictive value; PPV, positive predictive value; mGPS, modified Glasgow Prognostic Score; CRP : Alb, C-reactive protein to albumin ratio.

Table 4 Comparison of 90-day overall survival (90 days or less) in discovery cohort for each scoring system
Values are n (%); *percentage of cohort.Scottish malignant ureteric obstruction (MUO) Score and C-reactive protein to albumin ratio (CRP : Alb) are continuous scales, and are grouped by decile.PLaCT score and modified Glasgow Prognostic Score (mGPS) utilize three prognostic groups.