SUPPRESSION OF THE NITRIC OXIDE PATHWAY IN METASTATIC RENAL CELL CARCINOMA PATIENTS RECEIVING VEGF-SIGNALING INHIBITORS

Abstract

Therapies that target the vascular endothelial growth factor (VEGF) pathway cause hypertension but the mechanism remains unknown. This cross-sectional study tested the hypothesis that VEGF inhibition causes hypertension by suppressing VEGF-mediated vasodilatory pathways. Urine was collected from 80 patients with metastatic renal cell carcinoma from 2002–2009, 40 at baseline and 40 while on VEGF inhibitors. Measured urinary biomarkers include albumin, metabolites of the nitric oxide pathway and its downstream effector, cGMP, and prostaglandin pathway biomarkers prostaglandin E2, 6-keto PGF 1α, and cAMP, all normalized to urinary creatinine. The mean age in both groups was 61.8 years, 76% were male, and urinary albumin was higher in patients receiving VEGF inhibitors (median 18.4mg/g vs. 4.6 mg/g; p=0.009). cGMP/Cr was suppressed in patients on VEGF inhibitors (0.28 pmol/ug vs. 0.39 pmol/ug; p=0.01), with a trend toward suppression of nitrate/Cr (0.46 umol/mg vs. 0.62 umol/mg; p=0.09). Both comparisons were strengthened when patients on bevacizumab were excluded and only those receiving small molecule tyrosine kinase inhibitors were analyzed (cGMP/Cr, p=0.003; Nitrate/Cr, p=0.01). Prostaglandin E2, 6-keto PGF1α, and cAMP did not differ between groups. These results suggest that hypertension induced by VEGF inhibitors is mediated by suppression of nitric oxide production. Prospective studies are needed to explore whether these biomarkers may be useful predictors of efficacy in patients receiving VEGF-targeted therapies.

Introduction

The use of vascular endothelial growth factor (VEGF)-targeted therapies, also called antiangiogenic therapies, for treatment of solid tumors is growing rapidly.¹ FDA-approved antiangiogenic therapies include bevacizumab, a humanized monoclonal antibody directed against VEGF, and small molecule multi-targeted kinase inhibitors including sorafenib, sunitinib, and pazopanib. The small molecule angiangiogenic drugs target VEGF receptors as well as the platelet-derived growth factor receptor, RAS, and c-KIT.² Together, these new therapies have had a profound impact on the treatment of multiple tumor types.

VEGF-targeted therapies are strongly associated with development of both hypertension and proteinuria. Some studies have noted hypertension in over 80% of treated patients and an absolute elevation in blood pressure (BP) in 93% of patients on sorafenib.^3–6 Guidelines for the surveillance and management of hypertension in these patients were recently published by The National Cancer Institute but they note that the mechanism of hypertension remains undetermined and little data exists on which to base treatment recommendations.⁷ Proteinuria, measured as dipstick albuminuria, is also seen in 20–63% of patients, with a higher risk for renal cell carcinoma than for other tumor types.^8–10 Importantly, the development of hypertension on VEGF-targeted therapy correlates with improved cancer outcomes, and therefore could represent a biomarker useful for selecting the subset of patients with the best chance of responding to antiangiogenic therapy.^11–14 Understanding the pathophysiology of hypertension on these agents may therefore lead to the development of other novel efficacy biomarkers.

Preclinical evidence implicates the inhibition of VEGF signaling in development of hypertension on these agents. VEGF stimulates production of vasodilatory nitric oxide (NO) via activation of endothelial nitric oxide synthetase, ^{15, 16} and also upregulates vasodilatory prostacyclin. ^{17, 18} Small molecule VEGF-targeted therapies inhibit the nitrate pathway in vitro,¹⁹ and in humans, hypertension often develops within the first few days of treatment, consistent with acute suppression of NO- and/or prostacyclin-mediated vasodilation.^20–22 In mice, blockade of VEGF signaling by administration of an antibody directed against VEGFR-2 also caused hypertension but addition of an NO inhibitor abolished the blood pressure difference between control and antibody groups, suggesting that VEGF inhibition causes hypertension by depressing NO levels.²⁰ These observations led us to hypothesize that in humans, VEGF-targeted antiangiogenic therapies suppress the nitrate and prostaglandin pathways, leading to hypertension. This cross-sectional study explored urinary biomarkers of the nitrate and prostaglandin pathways in patients with renal cell carcinoma, comparing patients receiving antiangiogenic therapy with patients not receiving antiangiogenic therapy.

Methods

Study Population

Urine samples were collected from 300 patients in the Dana Farber Cancer Institute Kidney Cancer Center between 2002 and 2009. The 80 patients for this particular study were all patients with metastatic renal cell carcinoma for whom we had a single urine collection. Forty patients were on antiangiogenic therapy at the time of the urine collection and were all included in this analysis, and 40 control patients were selected by a random selection method amongst the 60 patients who went on to initiate first-line VEGF inhibitors within the 6 months after collection of the urine sample, since we did not have enough candidates for controls to match on clinical factors. We assume that those who went on the medications shortly after the urine collection would likely have had similar tumor characteristics to the 40 patients on the medications at the time of the collections.

Inclusion criteria for the study were a biopsy-confirmed diagnosis of metastatic renal cell carcinoma and the VEGF inhibitor criteria listed above. Exclusion criteria included the use of nitrate medications for heart disease and uncontrolled baseline blood pressure (BP) prior to starting VEGF-targeted therapy (systolic BP >180, diastolic BP >110).

Urine was collected by mid-stream clean catch at clinic visits and within four hours of collection it was centrifuged at 2,000 × g to sediment cellular debris. The supernatant was aliquoted without addition of protease inhibitors and stored at −80°C until analysis.

This study was approved by the institutional review board, and all patients signed consent forms for the urine sample collections for research use.

Measurement of Biomarkers

Commercially available kits were used to measure nitrates, the NO metabolite, and its downstream effector cyclic GMP(cGMP), as well as cyclic AMP(cAMP), prostaglandin E2 (PGE2), and 6-keto PGF 1α, a stable metabolite of prostacyclin (6-keto PGF 1α from Cayman Chemical, Ann Arbor, MI, all others from R&D Systems, Minneapolis, MN). All biomarker values were normalized to urinary creatinine to account for differences in urinary concentration. Nitrates were measured by fluorescence spectroscopy, and were measured as NO_x. NO_x is the sum of NO₂+NO_3, the stable metabolites of NO, after subtraction of endogenous nitrites. The CV for the assay was 3.5%. cGMP, cAMP, PGE2, and 6-keto PGF 1α were measured by ELISA, and the CVs for these assays were 9%, 8.2%, 12.4%, and 9% respectively. Albumin and creatinine were measured using an Afinion microalbumin machine. CVs for both of these assays were 3%. Albuminuria was calculated as the albumin:creatinine ratio (ACR), (albumin/creatinine)*100.

Measurement of Blood Pressure

BP was measured at the time of the urine collections for the majority of the participants. For those in which there was no BP recorded at the time of the urine collection, the BP from the closest recorded clinic visit was used. For patients on VEGF inhibitors, baseline BPs were recorded from the medical record as the BP immediately prior to the time at which they started the medication. Although BPs were measured at clinic visits, there was no standardized protocol for measurement of BPs in these patients. There was no difference in measurement technique between patients in the different groups. Because many patients receiving VEGF-targeted therapy were started on antihypertensive medications, we also recorded the number of BP medications each patient was taking at the time of the urine collection.

Analysis of Covariates

Clinical information was collected by medical record review by study co-investigators. Covariates included age, body mass index, prior hypertension, family history of hypertension or cardiovascular disease, nephrectomy status, current smoking, and estimated GFR by the Chronic Kidney Disease Epidemiology Collaboration equation, chosen instead of the Modification of Diet in Renal Disease equation because of increased accuracy with values of eGFR >60 ml/min per 1.73 m².²³

Statistical Analysis

Wilcoxon rank sum tests were used for linear comparisons between the two groups for albuminuria and for urinary biomarkers, which were not normally distributed, and t-tests were used for all other linear comparisons. Chi square tests were used for binary comparisons. Spearman correlation coefficients for non-normal data were used for correlations between ACR and other urinary biomarkers.

Nicotine is known to decrease urinary cGMP levels in non-smokers, but to increase urinary cGMP levels in smokers.²⁴ Therefore, it would be important to look at smokers and non-smokers separately. However, because there were not enough smokers for a separate analysis, we ran secondary analyses excluding smokers. We also performed secondary analyses with log-transformed linear regression models to determine whether certain covariates confounded the relationship between VEGF inhibitor use and cGMP/Cr, and two-variable logistic regression models to detect any impact of covariates affecting the association between cGMP/Cr and macroalbuminuria.

Additional analyses were performed comparing the urinary biomarkers in patients on bevacizumab separately from patients on small molecule inhibitors, as they have different mechanisms of action and therefore their effect on vasodilatory pathways might be different.

Results

Patient characteristics

The average age in both groups was 61.8 years, and 23% of the patients on VEGF inhibitors and 25% of those not on VEGF inhibitors were female (Table 1). This is reflective of the demographics of metastatic renal cell carcinoma. Baseline systolic BP was similar in both groups. Although the only statistically significant difference between the two groups was in the frequency of prior hypertension, there were more nephrectomies and higher angiotensin converting enzyme-inhibitors use in the VEGF inhibitor group.

Table 1.

Characteristics of the cohort

Clinical variable	VEGF inhibitor (N=40)	No VEGF inhibitor (N=40)	P value for difference
Age, years	61.8 (10.1)	61.8 (9.6)	0.98
Nephrectomy, %	83	68	0.12
Prior HTN, %	73	50	0.04
Diabetes, %	18	13	0.76
ACE-I at sample, %	36	23	0.21
Female, %	23	25	0.79
Current smoking, %	11	15	0.50
Baseline systolic BP, mmHg	126.7 (18.6)	124.7 (16.6)	0.49
Baseline diastolic BP, mmHg	74.3 (10.0)	71.5 (9.0)	0.52
eGFR at urine collection, ml/min/1.73m2	60.3 (17.3)	60.0 (18.7)	0.64
Days of treatment prior to collection*	88 (39–183)	N/A

HTN denotes hypertension; VEGF, vascular endothelial growth factor; ACE-I, Angiotensin converting enzyme inhibitor; BP, blood pressure.; eGFR, estimated glomerular filtration rate

Mean (SD) or % unless otherwise stated

^*Median (IQR)

Comparison of urinary biomarkers

Cyclic GMP was suppressed in patients on VEGF inhibitors (p=0.01), and nitrate levels also trended toward suppression in these patients, although this was not statistically significant at p=0.09. cAMP, PGE2, and 6-keto PGF 1α were not different between the two groups (Table 2). After removing smokers from the analysis, cGMP was still suppressed in patients on VEGF inhibitors (p=0.02). Because of the differences in nephrectomy status, angiotensin converting enzyme-inhibitor use, and prior hypertension between the two groups, secondary analyses were performed with log-transformed linear regression to assess whether VEGF inhibitor status predicted cGMP/Cr level independent of these variables. While nephrectomy status did confound the results, the association between cGMP/Cr and VEGF-targeted therapy remained statistically significant (Table 3).

Table 2.

Urinary biomarkers in renal cell carcinoma patients on and off VEGF inhibitors with median (IQR) and p value for comparison by Wilcoxon test

Biomarker	VEGF inhibitor (N=40)	Non- VEGF inhibitor controls (N=40)	P value
NOx /Cr, umol/mg	0.46 (0.31–0.71)	0.62 (0.42–0.84)	0.09
cGMP/Cr, pmol/ug	0.28 (0.21–0.39)	0.39 (0.30–0.62)	0.01
PGE2/Cr, pg/ug	1.20 (0.79–2.00)	1.31 (0.99–1.69)	0.67
cAMP/Cr, pmol/ug	5.05 (4.60–7.19)	5.12 (3.88–6.17)	0.25
6-keto PGF 1α/Cr, pg/ug	1.08 (0.87–1.59)	1.13 (0.75–1.55)	0.67
ACR, mg/g	18.4 (5.1–236.2)	4.6 (0–35.6)	0.009

VEGF denotes vascular endothelial growth factor; IQR, interquartile range; NO_x, nitric oxide; Cr, creatinine; cGMP, cyclic GMP; PGE2, prostaglandin E2; cAMP, cyclic AMP; ACR, albumin:creatinine ratio

Table 3.

Multivariate-adjusted model of the association between log-transformed urinary cGMP/Cr and VEGF inhibitor

Covariates included in model	β Coefficient for VEGF inhibitor use	p value for VEGF inhibitor use
VEGF inhibitor use	−0.30	0.018
VEGF inhibitor use + age	−0.30	0.014
VEGF inhibitor use, age, prior HTN, ACE-inhibitor	−0.32	0.013
VEGF inhibitor use, age, prior HTN, ACE-inhibitor, nephrectomy	−0.24	0.047

cGMP/Cr denotes cyclic GMP:creatinine ratio; VEGF, vascular endothelial growth factor; HTN, hypertension; ACE, angiotensin-converting enzyme

ACR values were significantly higher in patients receiving VEGF inhibitors (median 18.4 vs. 4.6; p=0.009). Of the 40 patients on VEGF inhibitors, 13 patients had microalbuminuria (17–250mg/g for men and 25–355 mg/g for women)²⁵ and 8 patients had macroalbuminuria (>250mg/g for men and >355mg/g for women)²⁵. Of the 40 patients not receiving VEGF inhibitors, 12 patients had microalbuminuria but only 1 patient had macroalbuminuria. The p values for the difference between the two groups for microalbuminuria and macroalbuminuria were p=1.0 and p=0.03 respectively.

Comparison between class of VEGF inhibitor

Bevacizumab acts by binding and sequestering soluble VEGF whereas the small molecule antiangiogenic therapies block intrinsic VEGF receptor tyrosine kinase activity. Because these separate mechanisms of action might influence vasodilatory pathways differently, we performed a stratified analysis and compared both groups to the non-VEGF inhibitor controls (Table 4). Comparing the patients receiving only small molecule inhibitors to patients not receiving any VEGF inhibitor strengthened the association between antiangiogenic therapy and suppression of cGMP/Cr and NO_x /Cr (p=0.003 for cGMP/Cr, p=0.01 for NO_x /Cr). However, there was no significant difference between these levels when comparing only patients on bevacizumab with controls and no differences in these associations for any of the other biomarkers. ACR values were higher in patients on both types of VEGF inhibitor compared to control patients.

Table 4.

Urinary biomarkers in renal cell carcinoma patients on small molecule VEGF inhibitors only (N=29) and bevacizumab only (N=11) compared with controls not on VEGF inhibitors with median (IQR) and p value for comparison by Wilcoxon test

Biomarker	Non- VEGF inhibitor controls (N=40)	Small molecule VEGF inhibitors (N=29)	P value vs. controls	Bevacizumab (N=11)	P value vs. controls
NOx /Cr, umol/mg	0.62 (0.42–0.84)	0.36 (0.26–0.58)	0.01	0.67 (0.45–1.41)	0.52
cGMP/Cr, pmol/ug	0.39 (0.30–0.62)	0.25 (0.22–0.36)	0.003	0.47 (0.19–0.77)	0.66
PGE2/Cr, pg/ug	1.31 (0.99–1.69)	1.08 (0.78–1.82)	0.49	1.50 (0.80–2.54)	0.78
cAMP/Cr, pmol/ug	5.12 (3.88–6.17)	5.09 (4.68–7.43)	0.15	4.71 (4.27–6.49)	0.18
6-keto PGF 1α/Cr, pg/ug	1.13 (0.75–1.55)	1.03 (0.85–1.31)	0.86	1.26 (0.94–2.22)	0.98
ACR, mg/g	4.6 (0–35.6)	18.5 (0–232.3)	0.03	18.3 (7.9–607.9)	0.04

VEGF denotes vascular endothelial growth factor; IQR, interquartile range; NO_x, nitric oxide; Cr, creatinine; cGMP, cyclic GMP; PGE2, prostaglandin E2; cAMP, cyclic AMP; ACR, albumin:creatinine ratio

We next compared urinary biomarkers in patients on the two classes of VEGF inhibitors (Table 5). NO_x /Cr levels were higher in patients on bevacizumab (0.67 vs. 0.36; p=0.01). There was no statistically significant difference between any of the other biomarkers, but given the low number of patients on bevacizumab we may lack power to detect a difference.

Table 5.

Comparison of urinary biomarkers and blood pressures in patients on bevacizumab and patients on small molecule VEGF inhibitors, with medians (IQR) and p values by Wilcoxon test for comparisons

Biomarker or clinical variable	Bevacizumab (N=11)	Small molecule VEGF inhibitors (N=29)	P value
NOx /Cr, umol/mg	0.67 (0.45–1.41)	0.36 (0.26–0.58)	0.01
cGMP/Cr, pmol/ug	0.47 (0.19–0.77)	0.25 (0.22–0.36)	0.28
PGE2/Cr, pg/ug	1.50 (0.80–2.54)	1.08 (0.78–1.82)	0.48
cAMP/Cr, pmol/ug	4.71 (4.27–6.49)	5.09 (4.68–7.43)	0.30
6-keto PGF 1α/Cr, pg/ug	1.26 (0.94–2.22)	1.03 (0.85–1.31)	0.07
ACR, mg/g	18.3 (7.9–607.9)	18.5 (0–232.3)	0.55
Number of blood pressure medications	2 (0–3)	1 (0–2)	0.82
Days of medication prior to collection	140 (80–245)	70 (21–168)	0.09

^*Mean (SD)

VEGF denotes vascular endothelial growth factor; IQR, interquartile range; NO_x, nitric oxide; Cr, creatinine; cGMP, cyclic GMP; PGE2, prostaglandin E2; cAMP, cyclic AMP; ACR, albumin:creatinine ratio

Blood pressure analysis

BP values at baseline between the patients on VEGF inhibitors and those not on VEGF inhibitors were similar (Table 1). After starting therapy, the mean systolic BP in patients on VEGF inhibitors was 131.8mmHg, compared with the mean systolic BP in patients not on VEGF inhibitors of 124.7mmHg (p=0.09). However, the median number of BP medications in patients receiving VEGF-targeted therapy was 1.5 (IQR 0–2.5), compared with a median of 0 (IQR 0–1) medications in patients not on VEGF-targeted therapy (p=0.002 for the comparison). There was no difference in the number of BP medications in patients on bevacizumab and those on other VEGF-targeted therapies (Table 4).

Analysis of albuminuria

Spearman correlation coefficients were calculated between ACR and all of the urinary biomarkers (Table 6). Correlations were noted between ACR and both cGMP/Cr (r=0.44, p=0.004) and 6-keto PGF 1α/Cr (r=0.31, p=0.05). Although the small number of outcomes limited our ability to do multivariate regression, bivariate regression revealed that even after adjustment for age, prior hypertension, nephrectomy status, diabetes, and angiotensin converting enzyme inhibitor use in logistic regression models, cGMP/Cr was significantly associated with macroalbuminuria (p<0.05 for all bivariate analyses).

Table 6.

Spearman correlation coefficients between biomarkers and ACR in patients on VEGF-targeted therapies (N=40)

Biomarkers	Spearman correlation coefficient	P value
NOx /Cr	−0.08	0.61
cGMP/Cr	0.44	0.004
PGE2/Cr	0.12	0.46
cAMP/Cr	0.07	0.66
6-keto PGF 1α/Cr	0.31	0.05

ACR denotes albumin:creatinine ratio; VEGF, vascular endothelial growth factor; NO_x, nitric oxide; Cr, creatinine; cGMP, cyclic GMP; PGE2, prostaglandin E2; cAMP, cyclic AMP; ACR, albumin:creatinine ratio

Discussion

In this cross-sectional pilot study, urinary biomarkers of the NO pathway were suppressed in patients receiving VEGF-targeted chemotherapies. Although the suppression of nitrate levels was not statistically significant, its measurement can be affected by diet and cGMP may be a more accurate reflection of NO pathway activity.²⁶

These findings remain significant after adjusting for age, prior hypertension, angiotensin converting enzyme-inhibitor use, and nephrectomy status, although nephrectomy status did change the effect estimate. As expected, PGE2 and cAMP were not influenced by VEGF inhibition. Although VEGF can regulate vasodilatory prostacyclin production, 6-keto PGF 1α was not suppressed in this study. Together, these results support the theory that hypertension associated with VEGF-targeted therapies is caused by inhibition of nitric oxide-mediated vasodilation.

These results are consistent with preclinical and clinical data that support a central role for NO in hypertension caused by VEGF-targeted therapies. Infused VEGF rapidly induced hypotension in an NO dependent fashion.^{20, 21, 27} Similarly, BP rises rapidly -- within 24 hours -- in patients who initiate therapy with VEGF inhibitors, possibly reflecting acute inhibition of vasodilation.⁴ VEGF inhibition may also contribute to hypertension by other mechanisms. For example, the proximal tubule natriuretic response to elevated blood pressure is partially dependent on cGMP and VEGF-targeted therapies might suppress this response, perpetuating the rise in blood pressure.^28–30 Our data do not rule out a contribution from capillary rarefaction to hypertension induced by VEGF blockade, as has been proposed,^{31, 32} or from increased circulating endothelin-1 as recently reported.³³

Although only 11/40 (28%) of patients were on bevacizumab and the rest were on small molecule VEGF receptor inhibitors, the difference in biomarkers between the two groups is striking. This is the first study reporting these comparisons, and inhibition of the NO pathway was much more profound in patients receiving small molecule VEGF inhibitors. Although not statistically significant, patients on bevacizumab had been in the study longer by the time of the urine collection (140 days vs. 70 days; p=0.09). However, they were similar with respect to prior hypertension (64% vs. 75%; p=0.44), nephrectomy status (73% vs. 86%; p-0.32), angiotensin coverting enzyme-inhibitor use (36% vs. 34%; p=0.82), diabetes (18% vs. 17%; p=0.94), and median ACR values (18.3mg/g vs. 18.5mg/g; p=0.55). The reason for these findings requires further investigation.

In both patients on bevacizumab and other types of VEGF inhibitors, ACR was elevated and there was a higher incidence of macroalbuminuria than in patients not on VEGF inhibitors. These results are expected because albuminuria is a well-described complication of antiangiogenic therapy reflecting inhibition of paracrine VEGF signaling between VEGF-producing glomerular podocytes and adjacent endothelial cells.³⁴ Inhibition of podocyte-endothelial cell VEGF signaling, whether through genetic or pharmacologic means, causes endotheliosis, thrombotic microangiopathy, and narrowing of the capillary lumen- the pathologic lesion seen in human kidney biopsy specimens from patients with albuminuria receiving VEGF-targeted therapies.^{34, 35}

Since endothelial knockout of NO leads to renal thrombotic microangiopathy in mice,³⁶ and albuminuria from chronic VEGF inhibition likely reflects renal thrombotic microangiopathy in humans,³⁵ we expected that patients receiving VEGF-targeted therapy with higher levels of albuminuria would also have suppressed NO pathway biomarkers. However, we observed that although NO pathway activity was lower than in control patients not receiving these drugs, urinary cGMP positively correlated with a higher degree of albuminuria, indicating that individuals with albuminuria had the least degree of nitrate pathway inhibition. The reasons for this observation are unclear. Acute thrombotic microangiopathy syndromes can be associated with an upregulation of endothelial NO synthetase in rats, potentially as a vasodilatory compensation mechanism to maintain renal perfusion.³⁷ Elevation of endothelial NO synthetase has been observed in renal biopsies of patients with diabetic nephropathy, a glomerular pathology characterized by endothelial damage.³⁸ Finally, increased endothelial shear stress can activate endothelial NO synthetase activity directly, presumably through a VEGF-independent mechanism.³⁹ Though clearly speculative, chronic VEGF inhibition may cause endothelial dysfunction with albuminuria with subsequent compensatory activation of endothelial NO signaling through VEGF-independent pathways.^{40, 41} The increase in 6-keto PGF 1α with increasing ACR (Spearman correlation coefficient r=0.31, p=0.05) also supports this theroay, as prostacyclins can be upregulated in the setting of endothelial damage as well.^{42, 43}

There are inherent limitations to this analysis. BP measurements were not standardized. However, there was no difference in the way BP was measured between the two groups or from baseline to the time of the urine collection in the VEGF inhibitor group. This was a cross-sectional study, comparing patients on and off VEGF inhibitors. While we tried to account for this by choosing patients for our control group only if they were started on VEGF inhibitors within 6 months of the collection, using these two groups instead of a prospective study is still not optimal. Urine specimens were collected from patients at different time points during their medication use. This may influence biomarkers if they change over time, and would bias our results to the null when compared with measurements over standardized time intervals.

Perspectives

This pilot study demonstrates NO pathway suppression in patients on VEGF-targeted therapies. The results suggest a difference in pathophysiology between the mechanisms of hypertension and albuminuria induced by VEGF inhibition. Understanding the pathophysiology of these toxicities takes on increased importance as the clinical use of VEGF-targeted therapies grows. Since blood pressure rise itself may represent a biomarker for clinical efficacy in individual patients, these results suggest that urinary cGMP might also be a useful efficacy biomarker less susceptible to the variations of office BP measurement.⁶ Future prospective studies will be required to correlate these novel biomarkers with cancer outcomes and the possible role of other pathways such as prostacyclins or endothelin-1.⁴⁴