.png)
A New Strategy For Blood Oxygen Level-Dependent MRI Assessment of Hypoxemia
Written by: Gerren Hobby, MD
Expert reviewer: Pottumarthi Prasad, PhD
AcademicCME (www.academiccme.com) is accrediting this educational activity for CE and CME for clinician learners. Please go to https://academiccme.com/kicr_blogposts/ to claim credit for participation.
INTRODUCTION
When talking about kidney disease, we often focus on glomerular disorders, but when it comes to the progression of chronic kidney disease, it's the combination of factors like a loss of parenchymal cells, chronic inflammation, diminished regenerative capacity of the kidney, and the final common path of fibrosis that pushes one towards dialysis. Glomerular injury leads to the loss of podocytes and the release of cytokines and growth factors. These, in turn, activate myofibroblasts responsible for producing the extracellular matrix (ECM). Nephrons are lost and replaced with scar tissue. This is the hallmark of chronic kidney disease progression. Multiple aspects of CKD progression are being studied, but interestingly kidney ischemia has been difficult to study or measure. Extracellular matrix (ECM) expansion increases the distance from blood vessels to tubule cells, resulting in hypoxia. Oxygen deprivation causes scarring of the glomerulus and interstitium, thereby driving CKD progression. Although its presence has long been suggested by the loss of peritubular capillaries on biopsies of CKD patients, firm proof of hypoxia in human kidneys has eluded us.
In the late 1990s, the chronic hypoxia hypothesis emerged and started with the fact that glomerular disease alters downstream blood flow in the peritubular capillaries. Altered gene activity and a cascade of other effects ensue, culminating in fibrosis. Hypoxia-inducible factors (HIF) comprise some of the key upregulated genes. As their activity increases, they alter the metabolism of tubular epithelial cells thereby causing extracellular matrix accumulation, interstitial collagen deposition, and matrix metalloproteinases suppression which normally breaks down said extracellular matrix.
The above processes provide a plausible mechanism for CKD progression from hypoxia. However, the question remains; does hypoxia actually exists in the kidneys of patients with chronic kidney disease? To begin, studies from animal models directly show that hypoxemia actually exists in animal models of chronic kidney disease. Utilizing oxygen microelectrodes, it was directly shown that hypoxia exists in CKD. For obvious reasons, these studies have not been performed in humans and because of this, our evidence base for the presence of hypoxia in humans with chronic kidney disease is more peripheral. Kidney biopsies of patients with chronic kidney disease reveal two important findings: rarefaction of peritubular capillaries and expansion of the extracellular matrix. These observations suggest a hypoxic environment. Additionally, studies using hypoxia-dependent pimonidazole protein adducts indicate hypoxia in CKD. Pimonidazole binds to thiol groups of protein, peptides, and amino acids at oxygen tensions below 10mmHg. Increased level of staining with this compound indicate lower oxygen levels. Pimonidazole protein adducts give more of a binary result though, and don’t allow us to quantify the degree of hypoxia. Lastly, proteomic studies show increased expression of HIF in patients with CKD.
In summary, we have concrete evidence in animals, and peripheral evidence in humans, to support the chronic hypoxia hypothesis. We have the biological plausibility of the role of hypoxia in fibrosis development, but we lack direct evidence of hypoxia in human chronic kidney disease patients.
What we need is a deeper understanding of hypoxia in human CKD to explore further the link between hypoxia, fibrosis, and chronic kidney disease. This brings us to the recent paper in Kidney International Reports, Quantitative Blood Oxygenation Level Dependent Magnetic Resonance Imaging for Estimating Intra-renal Oxygen Availability Demonstrates Kidneys Are Hypoxemic in Human CKD, which invites us into the world of magnetic resonance imaging (MRI).
We are all familiar with popular functional MRI studies that reveal changes in brain activity in response to a certain stimulus. Interestingly, these MRI studies, which utilize blood oxygen-dependent MRI (BOLD MRI), are an excellent tool to examine oxygen levels in CKD patients. The current KI Reports paper does just that, and provides the crucial missing piece of information we need for the hypoxia hypothesis.
HOW MRI WORKS
This paper is laden with technical details, so it’s worthwhile to spend some time discussing the physics of MRI. Figure 1 below is an overview of MRI physics, but also, be sure to check out this amazing video that gives a great introduction. You can find a more in-depth introduction to basic MRI physics in this video, as well as this one.
Figure 1: A: Protons are the basis of MRI imaging and are mainly found in water and fat in the body. They behave like small bar magnets, having one north pole and one south pole. In their normal state, their poles face different directions, and this random orientation cancels out their individual magnetization to create a net magnetization vector of zero. In addition, protons spin and their action of spinning is termed “precession”. Normally, protons spin (precess) out of sync with one another in a random fashion. B: The powerful fixed magnet of MRI aligns the poles of the protons along its Z axis, either in a parallel or antiparallel fashion, with a small majority aligning with the field of the MRI machine. Even though the fixed magnet aligns the protons along the Z axis, the second feature remains unchanged – they still spin (precess) out of sync with one another. C: When an MRI image is obtained, a radiofrequency (RF) pulse is applied at a 90-degree angle which tips the poles of the protons from the Z-axis 90 degrees into the X-axis. In addition, it causes the protons to spin (precess) in sync. D: When the RF pulse is stopped, the protons relax back to the Z-axis and they begin to relax and precess out of sync with one another.
When the RF pulse is removed, as shown in panel D, it creates what we think of as longitudinal relaxation (T1 relaxation) and transverse relaxation (T2 relaxation) which are due to regrowth of the longitudinal and decay of the transverse signal, respectively.
T1 relaxation is caused by protons relaxing from their X-axis orientation back to the longitudinal Z-axis
T2 relaxation is a bit more complicated. The transverse signal is maintained during an RF pulse when protons precess in sync with one another in addition to orienting their poles in the X-axis. Any loss in the synchronicity of proton precession or a return of the poles to the Z axis results in decay of the transverse signal and thus T2 relaxation.
Different body tissues have different T1 and T2 relaxation times and this forms the basis of MRI imaging and how images of different organs are formed. Placing an emphasis on the differing T1 or T2 relaxation times also forms the basis of T1- or T2-weighted imaging.
The current paper in KI reports utilizes an MRI parameter called R2*. As mentioned above, T2 is the time it takes for decay of the transverse signal. R2 refers to the relaxation rate and is simply the inverse (1/T2) of the T2 signal. A short T2 time provides a smaller denominator for R2 and thus a higher number – a faster relaxation rate. Conversely, a long T2 relaxation time gives R2 a large denominator and a smaller R2 value. The R2 relaxation rate is determined by the strength of the MRI fixed magnet as well as the intrinsic properties of the proton itself. In general, as the magnetic field increases, the transverse signal decays faster and the R2 relaxation rate increases. A strong fixed magnet pulls the protons back to their original orientation faster than a weak magnet does. The reality is, of course, more complicated. In addition to the fixed MRI magnet, any substances in the body with magnetic properties will increase the local magnetization of protons and increase the relaxation rate (R2). Deoxyhemoglobin, as opposed to oxyhemoglobin, contains unpaired electrons and displays these paramagnetic properties. The local magnetic field in a tissue increases along with increasing deoxyhemoglobin, and this speeds up the relaxation rate (R2). This true, or “observed relaxation rate (R2)” is termed R2*. For the purposes of understanding this paper, the higher the R2*, the higher percentage of deoxyhemoglobin we will find in the tissue, suggesting tissue hypoxia. By analyzing the overall amount of deoxyhemoglobin (via the R2* parameter) within a standardized volume of kidney tissue, called a voxel, we can estimate oxygenation levels on a spatial level in the kidney.
This is an incredible technique, but unfortunately, past studies utilizing BOLD MRI in CKD have revealed conflicting results due to the limitations of this technique. Firstly, biopsies in chronic kidney disease show rarefaction of the microvasculature, which reduces the total blood volume in the kidney. Secondly, we frequently encounter anemia in CKD patients. These two factors reduce the total amount of hemoglobin molecules within a voxel through pericapillary rarefaction and reduction in the concentration of hemoglobin in blood.
Past studies of BOLD MRI used R2* to estimate oxygen levels in the kidney. However, as stated above, both pericapillary rarefaction and anemia lower the total amount of hemoglobin (and thus paramagnetic deoxyhemoglobin that MRI detects) in the volume of a voxel. This led to lower R2* levels which gave false impressions of higher oxygenation labels in the kidney. This assumption is due to the incorrect conclusion that if there are low levels of deoxyhemoglobin present, then there must be high levels of oxyhemoglobin. This gave the false impression of higher oxygen levels in CKD patients, thus this interpretation error led to conflicting results in the previous BOLD MRI literature in CKD patients. The current study, however, takes an intriguing approach by incorporating a contrast agent to measure the overall blood volume, thus removing the variable of pericapillary rarefaction in the kidney. Additionally, they measured the hematocrit from peripheral blood samples, taking into account the only other variable that adversely affects R2* for proper estimation of oxygenation in the kidney.
This study utilizes the IV iron medication, ferumoxytol to estimate blood volume. Why would they do this? It has parametric properties (order of magnitude higher strength of deoxyhemoglobin in magnetic properties) and thus increases the R2* signal. When injected, it goes into the bloodstream, and increases the R2* signal. Utilizing pre- and post-ferumoxytol imaging, the fractional blood volume of the kidney can be estimated.
METHODS
This study evaluated kidney oxygenation in CKD patients and healthy controls. BOLD MRI was used to image the kidneys of 9 healthy controls and 6 individuals with CKD and utilized R2* as the parameter to estimate oxygenation in the kidney.
Inclusion Criteria for all patients were as follows:
Age ≥ 18 years
Ability to give informed consent and willingness to follow study protocol
Exclusion criteria for all participants included:
Contraindication for MRI
Pregnant or nursing females
History of heart failure
History of renal artery stenosis
History of ureteral obstruction
NSAID use
Iron overload (i.e. ferritin >800 ng/mL
For CKD group: history of primary glomerular disease, primary interstitial disease, or polycystic kidney disease
Healthy participants had no history of HTN, DM, heart disease, or CKD. The CKD group had type 1 or type 2 diabetes and CKD G3-4.
MRI scans were completed with a 3.0 T MRI machine. qBOLD MRI data included both R2 and R2* measurements. After the initial R2* images were acquired, ferumoxytol was administered intravenously for estimation of blood volume.
RESULTS
Baseline characteristics of the participants are shown in Table 1.
When healthy controls and CKD patients were compared, we can see that the paramagnetic properties increased R2* values in the kidneys of both cohorts as expected (figure 1). After injection of ferumoxytol, the R2* increase is higher in healthy controls than CKD patients, which suggests a higher fractional blood volume in healthy controls.
When fractional blood volume is taken into account, it was found that the cortex is normoxic in healthy controls, but moderately hypoxemic in CKD. Next, the medulla was mildly hypoxemic in healthy controls and moderately hypoxemic in CKD (Figure 3).
DISCUSSION
This study gives us the first direct evidence of hypoxemia in human CKD. Since the chronic hypoxia hypothesis emerged, a number of pathogenic mechanisms connected hypoxia to the development of chronic kidney disease. Direct evidence from animal studies supported the presence of hypoxemia in CKD. Multiple layers of peripheral evidence also emerged from human studies, but we still lacked direct observations that hypoxemia existed in human CKD kidneys. Past studies utilizing BOLD MRI to examine hypoxemia revealed conflicting results due to methodological flaws. This study represents a significant step forward by using ferumoxytol to estimate blood volume to account for variables that affected oxygen estimation with the R2* parameter. There are a number of potential applications for this perfected technology. We now find ourselves in an era of nephrology with an increasing number of tools to affect the progression of chronic kidney disease. Testing the effect of a particular compound on kidney oxygenation could be one possible application. SGLT2 inhibitors slow CKD progression via mechanisms not fully explained by glucose lowering, blood pressure management, or a reduction in intraglomerular pressure. They have been shown to ameliorate hypoxia in the kidney and promote oxygen deprivation signaling, and this may partially mediate their kidney-protective effects. The utilization of BOLD MRI to fully understand these effects could be very useful. The authors additionally mentioned evaluation of oxygenation in the setting of renal artery stenosis, to determine which patients might benefit from interventions. In short, this is a tool that will be very helpful in studying chronic kidney disease, and it will be an exciting area of research to follow in the coming years.
AcademicCME (www.academiccme.com) is accrediting this educational activity for CE and CME for clinician learners. Please go to https://academiccme.com/kicr_blogposts/ to claim credit for participation.