PATIENTS AND METHODS
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
Following Institutional Review Board approval, the electronic medical records database was searched to identify adult patients who had undergone partial or radical nephrectomy for suspected RCC at the Duke University Medical Center between 2000 and 2008. Patients treated with ablative techniques were excluded. Similarly, we excluded cases with metastatic disease on presentation. We retrieved the following data: patient age, gender and race, personal and family (first-degree relatives) history of cancer, presentation (incidental versus symptomatic), lesion size on imaging expressed as the largest dimension, lesion morphology on imaging (solid versus complex cyst), pathological data regarding histological diagnosis, subtype, Fuhrman grade , extracapsular extension and vascular invasion. Incidental presentation was defined as asymptomatic lesions, discovered incidentally upon imaging for unrelated complaints.
Fuhrman grade was recorded for all RCC cases. Whenever two nuclear grades were reported, the higher one was used in this analysis. Whenever several radiographic studies were available, the most recent measurement before surgery was used based on contrast-enhanced CT or MRI.
We excluded from our analysis patients presenting with multiple lesions, as well as patients with Von Hippel–Lindau syndrome and other known pathologies (e.g. tuberous sclerosis) that may suggest the nature of the renal lesion. We established a cohort of patients undergoing surgery for an enhancing kidney lesion suspected to be RCC.
Pathological findings were described according to lesion size and logistic regression models were used to assess potential predictors of malignant pathology, adverse pathological features (high Fuhrman grade, vascular invasion and extracapsular extension) as well as predictors of potentially aggressive versus probably indolent kidney masses. Potentially aggressive disease was defined as at least one of the following characteristics on final pathology: non-RCC malignancy (sarcoma, squamous or transitional cell carcinoma, lymphoma, nephroblastoma), undifferentiated RCC, rare subtypes (medullary and collecting duct carcinoma), sarcomatoid/rhabdoid features, vascular invasion, extracapsular extension and high (3–4) Fuhrman grade.
For regression models, age and radiographic size were treated as parametric variables; all other considered variables were treated categorically. P values <0.05 were considered significant. Data are presented as mean (±sd), median (interquartile range) or number (%) unless otherwise specified. Statistical analysis was performed using SPSS v17 software (SPSS Inc., Chicago, IL).
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
A total of 768 records were identified that matched the criteria for this study. Patient demographics are reported in Table 1. Median tumour size on imaging was 4 cm (interquartile range 2.6–7.0). Malignancy was pathologically confirmed in 628 (81.8%) specimens, whereas no malignancy was found in the remaining 140 (18.2%). Malignant findings included RCC in 615 (97.9%) specimens and other (transitional cell carcinoma, squamous cell carcinoma, lymphoma, nephroblastoma and sarcoma) in the remaining 13 (2.1%). RCC histological subtypes were distributed as follows: clear cell in 466 (75.8%), papillary in 118 (19.2%), chromophobe in 20 (3.3%), collecting duct carcinoma in six (1.0%), medullary in one (0.2%) and unclassified in the remaining four (0.7%).
Table 1. Patient and tumour characteristics
| ||N (%)||Mean (SD)||Median (IQR)|
|Age (years)|| ||59.8 (12.0)||60.0 (52.0–67.8)|
|Gender:|| || || |
| Male||450 (58.6)|| || |
| Female||318 (41.4)|| || |
|Race|| || || |
| White||553 (72.0)|| || |
| Black||187 (24.3)|| || |
| Other|| 28 (3.7)|| || |
|Lesion size on imaging (cm)|| || 5.24 (3.67)|| 4.0 (2.6–7.0)|
|Radiographic appearance|| || || |
| Solid||669 (87.1)|| || |
| Bosniak III|| 66 (8.6)|| || |
| Bosniak IV|| 33 (4.3)|| || |
|Side|| || || |
| Left||377 (49.1)|| || |
| Right||391 (50.9)|| || |
|Personal history of malignancy|| || || |
| Negative||604 (78.6)|| || |
| Positive||164 (21.4)|| || |
|Family history of malignancy|| || || |
| Not available||213 (27.7)|| || |
| Negative||305 (39.7)|| || |
| Kidney|| 30 (3.9)|| || |
| Other than kidney||220 (28.6)|| || |
|Malignant pathology||628 (81.8)|| || |
| RCC||615 (97.9)|| || |
| Other|| 13 (2.1)|| || |
|Fuhrman grade (RCC)|| || || |
| 1|| 116 (18.9)|| || |
| 2||363 (59.0)|| || |
| 3||103 (16.7)|| || |
| 4|| 32 (5.2)|| || |
| Not applicable*|| 1 (0.2)|| || |
A total of 140 findings with no malignancy were distributed as follows: oncocytoma in 52 (37.1%), angiomyolipoma in 29 (20.7%), benign cyst in 26 (18.6%), cystic nephroma in nine (6.4%), infarct/necrotic debris in six (4.3%), haemorrhagic cyst in five (3.6%) and other (xanthogranulomatous pyelonephritis, metanephric adenoma, scar tissue, leiomyoma, myolipoma and normal kidney tissue) in the remaining 13 specimens (9.3%).
Pathology distribution according to radiological size of the lesion is reported in Table 2 and Fuhrman grade distribution is shown in Figure 1. Interestingly, benign findings accounted for more than one-third of lesions measuring <2 cm on preoperative imaging, decreasing progressively with increasing radiographic size. Renal cell carcinoma was the only malignancy found in smaller (<4 cm) tumours; however beyond 4 cm the incidence of malignant findings other than RCC (mainly sarcomas) increased, progressing from 2.0% for lesions 4.0-5.99 cm to 6.3% of tumours measuring >10 cm in the largest dimension.
Table 2. Final pathology distribution and radiographic lesion size
|Pathology N (%)||Radiographic size (cm)|
|Malignant||52 (64.2)||117 (76.0)|| 107 (82.3)||126 (84.0)||143 (90.5)||83 (87.4)||628 (81.8)|
| RCC||52 (64.2)||117 (76.0)|| 107 (82.3)||123 (82.0)||139 (88.0)||77 (81.1)||615 (80.1)|
| Other|| 0|| 0|| 0|| 3 (2.0)|| 4 (2.5)|| 6 (6.3)|| 13 (1.7)|
|Benign||29 (35.8)|| 37 (24.0)|| 23 (17.7)|| 24 (16.0)|| 15 (9.5)||12 (12.6)||140 (18.2)|
| ONC|| 5 (6.2)|| 14 (9.1)|| 12 (9.2)|| 11 (7.3)|| 9 (5.7)|| 1 (1.1)|| 52 (6.8)|
| AML||10 (12.3)|| 11 (7.1)|| 2 (1.5)|| 3 (2.0)|| 0|| 3 (3.2)|| 29 (3.8)|
| BC|| 6 (7.4)|| 6 (3.9)|| 5 (3.9)|| 6 (4.0)|| 1 (0.6)|| 1 (1.1)|| 25 (3.3)|
| Other|| 8 (9.9)|| 6 (3.9)|| 4 (3.1)|| 4 (2.7)|| 5 (3.2)|| 7 (7.4)|| 34 (4.4)|
Figure 1. Distribution of renal cell carcinoma Fuhrman grade according to preoperative radiographic size of the lesion. Note: Fuhrman grade was not applied to one case of unclassified subtype RCC specimen.
Download figure to PowerPoint
Table 2 illustrates the changing incidence of clinically and radiographically unsuspected angiomyolipoma (AML) based on tumour size, from 12.3% in the <2 cm category to 1.5–3.2% in lesions measuring ≥3 cm. A similar pattern was observed for the incidence of simple cysts (by final pathology) suspected of being RCC; the incidence was 7.4% in <2 cm lesions and progressively declined with increase of tumour size on imaging, becoming anecdotal in lesions ≥6 cm.
Binary logistic regression results for predicting malignant pathology, high Fuhrman grade, extracapsular extension and vascular invasion are shown in Table 3. Radiographic size was significantly associated with malignancy (versus benign pathology; odds ratio (OR) = 1.13, P= 0.001), high Fuhrman grade (OR = 1.21, P < 0.0001), vascular invasion (OR = 1.19, P < 0.0001) and extracapsular extension (OR = 1.23, P < 0.0001). Age was a significant predictor of high Fuhrman grade (OR = 1.02, P= 0.038) and vascular invasion (OR = 1.03, P= 0.018) with a trend toward extracapsular extension (OR = 1.02, P= 0.06). Male gender was associated with malignant pathology (OR = 1.73, P= 0.006), vascular invasion (OR = 2.33, P= 0.005), but not high Fuhrman grade or extracapsular extension. Solid radiographic appearance of the lesion was a significant predictor of malignancy (OR = 3.37, P < 0.0001) and vascular invasion (OR = 7.38, P= 0.012) but was not associated with high-grade or extracapsular extension. Positive family history of kidney cancer was associated with extracapsular extension but not with malignant pathology or other adverse characteristics.
Table 3. Binary logistic regression results to predict malignant pathology, high (3–4) Fuhrman grade, vascular invasion and extracapsular extension
| ||Malignant pathology||High grade RCC||Vascular invasion||Extracapsular extension|
|P value||OR||95% CI||P value||OR||95% CI||P value||OR||95% CI||P value||OR||95% CI|
|Race||0.489|| || ||0.426|| || ||0.428|| || ||0.585|| || |
| White (ref)||–||–||–||–||–||–||–||–||–||–||–||–|
|Hx of cancer||0.862||0.96||0.59–1.55||0.893||1.04||0.61–1.75||0.035||0.44||0.21–0.95||0.261||1.41||0.78–2.56|
|Family Hx of cancer||0.603|| || ||0.657|| || ||0.503|| || ||0.204|| || |
| Not available (ref)||–||–||–||–||–||–||–||–||–||–||–||–|
| Other than kidney||0.470||0.83||0.49–1.39||0.510||0.83||0.47–1.45||0.416||0.75||0.38–1.49||0.571||1.21||0.63–2.34|
Logistic regression results predicting potentially aggressive kidney masses are reported in Table 4. Age, symptomatic presentation, solid appearance and radiographic size were independent predictors of potentially aggressive disease, whereas for male gender (OR = 1.43, P= 0.062) a trend toward statistical significance was noted.
Table 4. Binary logistic regression results to predict aggressive disease defined as pathological findings of at least one of the following: non-RCC malignancy, high (3–4) grade RCC, medullary or collecting duct carcinoma, vascular invasion, extracapsular extension
| ||P value||OR||95% CI|
|Race||0.640|| || |
|Hx of cancer||0.989||1.00||0.64–1.57|
|Family Hx of cancer||0.186|| || |
| Not available||–||–||–|
| Other than kidney||0.651||0.89||0.55–1.45|
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
Widespread use and availability of abdominal imaging have changed the landscape of kidney lesions. Currently, enhancing renal lesions are usually serendipitously detected at an early stage and their incidence on autopsy has decreased . With a shift towards smaller lesions at presentation, contemporary treatment schemes may need reconsideration [4,8]. We analysed the association between radiographic size of kidney lesions and their pathological features in a contemporary series of patients undergoing surgery and assessed preoperative factors associated with potentially aggressive pathological findings.
We chose to base this study on radiographic size for several reasons. First, radiographic size has been shown to correlate well with the size reported by pathology . Second, radiographic size is the available parameter that is used for clinical decision-making and treatment planning. Finally, radiological size may be available for all enhancing lesions, whereas pathologists may not be able to report dimensions when no identifiable lesion is seen, such as in the case of pyelonephritis or ‘normal kidney tissue’ reports. Cases of known or highly probable pathology, such as patients with von Hippel–Lindau syndrome or tuberous sclerosis complex, or the few patients that had undergone a diagnostic biopsy of the lesion, were excluded to compose a cohort with no exceptional a priori probabilities of final diagnosis.
In this study, the incidence of enhancing lesions radiographically suspected to be RCC and subsequently identified as pathologically benign differed based on the size of the lesion on imaging. Very small (<2 cm) enhancing lesions undergoing surgery were benign in 36% of cases. This finding is similar to the recent observation of Thompson et al. that reported benign pathology in 37.5% and 19.2% of kidney lesions <1 cm and 1–2 cm in size, respectively, and correlated tumour size to malignant potential. Remzi et al. presented data on 287 tumours ≤4 cm whereby very small tumours (≤2 cm) were benign in 24.6% of cases. Frank et al. showed a similarly high incidence of benign lesions among small renal masses, reaching almost 50% in tumours <1 cm in size. Schlomer et al. found that 27.9% of kidney tumours <2 cm were benign. These authors performed their analyses with a ‘pathology-based’ approach by analysing records with histological evidence of tumour in the specimen, whereas we analysed all records with a radiographic finding of an enhancing lesion. Therefore, we were able to report on the ‘no pathologically identified benign or malignant tumour’ category including simple cyst, pyelonephritis and other findings probably not captured in the analyses of others. Discrepancies in the incidence of benign pathology between their data and our results may derive from this difference in analytical approach. If we censor the cases with no pathologically identified tumour, the malignant to benign ratio would become very similar across studies. However, we believe that all pathological entities that result in treatment must be considered when making clinical decisions based on radiological studies.
Fuhrman grade is an important prognostic factor in RCC. Although Fuhrman grade is associated with tumour size and stage it remains a strong independent prognostic factor . We found that the incidence of high-grade RCC is low among these very small tumours. Only five high-grade RCC (6%) were found among tumours measuring <2 cm. These findings are in accordance with other authors presenting similar results. Remzi et al. reported over 95% of tumours measuring ≤2 cm were low-grade (Fuhrman 1–2). Thompson et al. found that renal tumours <2 cm are low-grade in 90% of specimens. Similarly, in an earlier series, Frank et al. showed that small renal tumours are low-grade in >90% of cases. This finding argues for the relative indolence of these small tumours that may be adequately treated with less invasive methods such as needle ablative techniques, or expectantly followed, especially considering the higher incidence of benign pathological findings.
We found the incidence of AML to be surprisingly high (12.3%) among lesions measuring <2 cm, and 7.1% in lesions 2–2.99 cm. In larger enhancing lesions (≥3 cm) the incidence of AML was 0–3.2%. The finding is of interest because this mesenchymal tumour usually presents with a distinct radiographic appearance that may enable a correct preoperative diagnosis and direct the treatment strategy away from an aggressive surgical approach. Characteristic radiological findings of AML include area(s) of adipose tissue that may differentiate this renal lesion from RCC . It may be difficult, if not impossible, to perceive these typical features in a very small lesion with current imaging modalities, and this may account for the high incidence of AML in small lesions in our series. If these tumours were followed by serial tomography instead, subsequent imaging may reveal a previously ‘too small to characterize’ AML and could have avoided aggressive treatment in these patients.
It remains unclear whether biopsy of the suspected lesion could play a role in clinical management of these lesions. Diagnostic properties of percutaneous renal biopsies vary widely [16–19] although some report encouraging results, and complications may occur as a result of the procedure. Major problems that limit the clinical practice of kidney biopsies are the rate of non-diagnostic biopsy and the potential complications that have been recently reported to be around 10% each . At the Duke University Medical Center, percutaneous biopsies of a suspected renal lesion are not performed routinely, and we excluded from the current analysis patients who had undergone a diagnostic biopsy.
Regression analyses were performed to assess potential predictors of malignancy upon final pathology. We found that each 1 cm increase in radiographic size increases by 13% the chances of finding malignancy upon pathological evaluation of the surgical specimen. Solid lesions are three times more likely to be malignant. Radiographic size was strongly and independently associated with adverse pathological features of malignant masses with each 1 cm increase in size corresponding to roughly a 20% increase in the chance of high Fuhrman grade, vascular invasion or extracapsular extension. The importance of radiographic size in predicting the likelihood of malignant pathology is in agreement with the findings of other authors [10–13,20]. The independent effect of lesion dimensions on Fuhrman grade of the specimen is similar to other reports [10–12,20]. However, our regression models assessed the impact of radiographic size independently from other factors such as age and gender and confirmed its importance. Moreover, our findings show the independent association between radiographic size and other aggressive pathological features such as vascular invasion and extracapsular extension. In the preoperative nomogram proposed by Lane et al. to predict malignant disease, the results were similar because radiographic size was among the strongest factors and was included in the nomogram to predict potentially aggressive disease.
In the present study we also assessed preoperative risk factors for potentially aggressive disease. Our results suggest that radiographic size is among the strongest predictors with 27% increase in the chance of aggressive disease with each 1 cm increase in radiographic size. For example, a 6-cm lesion is twice as likely to harbour potentially aggressive disease compared with a 2-cm one. Other factors significantly associated with potentially aggressive disease are age and symptomatic presentation. By logistic regression, there are also notable trends for a positive association between male gender and family history of kidney cancer with the risk of potentially aggressive disease. These factors were included in the proposed predictive tool by Lane et al. but were not significantly associated with potentially aggressive disease on multivariate analysis and have probably contributed to overall suboptimal performance (C index = 0.557) of their nomogram. We did not attempt to validate their tool in our dataset because of discrepancies in the available variables.
The present study suggests that several preoperative variables are associated with pathological features of the renal mass. One of the strongest predictors of both malignant pathology and aggressive pathological features of the lesion is represented by radiographic size. These data may be helpful in counselling and clinical decision-making, and may provide additional information to identify candidates for less aggressive treatments based on the probability of the probably indolent pathology of the renal mass. Appropriate candidates with low risk of potentially aggressive pathology may be offered conservative management such as active surveillance that may delay or avoid interventional treatment.
There are several limitations to this study. First, this is a select cohort of patients who underwent surgery in a tertiary referral centre, therefore the results may not be applicable to different patient populations. Second, all the limitations of retrospective single-centre series apply to our study and the data should be interpreted accordingly, considering the possible biases. Third, we did not analyse the follow-up data because this is a contemporary cohort and such data would have been limited; nor did we include patients managed with ablative techniques or followed expectantly to obtain uniform pathological data. Finally, we did not perform a review of the specimens but relied on the original reports by our uropathologists, who are expert in evaluating kidney malignancies.
Despite the aforementioned limitations, the results of the present study suggest that enhancing sporadic kidney lesions measuring <2 cm on cross-sectional imaging are represented by benign lesions in more than one-third of cases. The remaining lesions are represented by RCC of predominantly low (1–2) Fuhrman grade. The data suggest the need for reappraisal of the current treatment strategy for small renal masses. Less aggressive approaches (compared with conventional surgery) such as needle thermoablative techniques or active surveillance may provide adequate tumour control with less morbidity in appropriately selected patients [21,22]. The results of this study may be useful in selecting candidates for active surveillance. In daily clinical practice these results may be of service in counselling patients with an incidentally discovered small enhancing renal lesion, although lesion size should not be the only parameter considered.
In conclusion, our experience reveals that solitary kidney lesions <2 cm suspected to be RCC are found to be malignant by final pathology in only 64% of cases. Additionally, these small-sized RCCs are mostly of low (1–2) Fuhrman grade. Our data suggest that age, male gender, radiographic size and appearance of the lesion, as well as symptomatic presentation, are associated with an increased risk of potentially aggressive disease. Hence, older, especially male, patients with solid larger lesions who present with symptoms that are probably related to the mass are likely to harbour potentially aggressive malignancy; whereas younger, female, patients with small kidney lesions that were discovered incidentally are more likely to have lesions with indolent pathology and may represent better candidates for surveillance and ablative techniques.