TARGET AUDIENCE Basics of cardiac MRI assumed known.

OBJECTIVES

Describe diffuse fibrosis

Understand myocardial T1 before and after Gad.

Connect to partition coefficient and myocardial extracellular volume fraction (ECV).

Recognise limits in acquisition and processing.

Learn realistic clinical research aims in diffuse fibrosis.

Describe improvements or possible other methods (even outside MRI).

Challenge: cardiac MRI not yet ready for individual diagnosis or monitoring of early-stage diffuse fibrosis.

GBCA = generic Gadolinium Based Contrast Agent GBCA , usually at 0.1-0.2mmol/kg dose.

“Gad” in the talk is just an easier way to say GBCA.

[Gd] = invivo concentration of Gad (millimoles/litre or milliMolar or mM).

Native = Before injection of Gad.

T1 = Longitudinal relaxation recovery time.

R1 = 1/T1 = Longitudinal relaxation rate (/s).

r1 = Longitudinal relaxivity of GBCA = "Power" of GBCA molecules to shorten water T1. (/s /mM ).

Relaxivity is defined by (1/T1) = (1/NativeT1) + r1 x [Gd] , same as R1 = NativeR1 + r1 x [Gd]

λ = Partition coefficient

ECV = Extracellular volume fraction of the myocardium (in the region of 0.3, or 30 if expressed as %).

Fractions of myocardial volume:

extravascular = outside any blood vessels (capillaries, arteries & veins)

intravascular = within any blood vessels

extracellular = outside cells

intracellular = inside cells

intercellular = interstitial = between myocardial cells such as cardiomyocytes.

Collagen is the flexible skeleton of myocardium.

Continual replacement → potential for rapid imbalance of myocardial collagen (Bishop et al 1995)

If dispersed throughout myocardium, such fibrosis cannot be detected by late-Gad = “diffuse fibrosis”.

Biomarkers in blood arise from any collagen synthesis, but most are non-specific to myocardium (Lopez et al 2015, Zannad 2014).

Histology: “picrosirius red” collagen stain , or Masson trichrome (blue collagen) shows collagen volume fraction (CVF). ECV is not directly seen in histology which can mislead due to tissue processing (fixing etc)

Elevated diffuse fibrosis increases myocardial stiffness → impaired diastolic relaxation (strain by MRE (Elgeti et al 2014); DENSE /early diastolic filling E-wave)

MRI does not normally receive collagen signal, nor the water in fast-exchange with collagen. Alternatives to T1&ECV? Short T2 ultrashort TE (UTE) / magnetization transfer contrast (MTC) / T1rho have been investigated and currently seem to be weak effects in myocardium compared to T1/ECV work.

Strong clinical calls for an early “subclinical” test for diffuse fibrosis as currently none except biopsy and diffuse fibrosis if detected early may be reversible without damage (Jellis et al 2010; Schelbert et al 2015).

This call is driving work to develop other diagnostics such as blood biomarkers if specific enough, and also in cardiac ultrasound.

GBCA(“Gad”) does not enter healthy intact cells.

This talk omits intravascular GBCA which do not enter interstitial space or do so very slowly.

Excluding conditions such as iron-loading or abnormal lipid content (e.g. Anderson-Fabry disease)

Pre-Gad ( = Native ) T1 increases with increasing pure water content of tissue.

Interstitial fluid is purer water than cytoplasm within cells.

Longer native T1 → increased (interstitial fluid / intracellular fluid) fraction.

Post-Gad T1 decreases with increasing Gad content of myocardium in the interstitial space.

Interstitial fluid contains Gad unlike cytoplasm within cells.

Shorter post-Gad T1 → increased fraction (interstitial fluid /intracellular fluid) i.e. reverse of native T1!

Assumption: Myocardium shows a single T1 (i.e. "monoexponential recovery curve") because of “fast transcytolemmal water-exchange” across cell walls. "Fast" is defined in comparison to the T1 and the relevant T1-weighting intervals of the MRI measurement method used. So cytoplasm and interstitial fluid are a “single compartment”, this is assumed to hold true when Gad shortens T1. (Donahue et al 1994&1997; Coelho-Filho et al 2013; Goldfarb et al 2016)

Similar “fast-exchange” is assumed in blood.

In most T1 measurement methods, myocardial contraction& relaxation occurs during the T1-weighting interval of the sequence method → there is interstitial fluid “microflow” during this time, causes more mixing than thermal diffusion alone would provide.

Assumption: Interstitial [Gd] = Capillary [Gd] = Left ventricular blood [Gd].

Originally performed by steady infusion. In non-infarcted myocardium, post bolus injection imaging late (>=15mins depending on dose, GBCA type, field strength) has been shown equivalent to steady infusion (European Society of Cardiology/SCMR consensus Moon et al 2013).

Assumption: Relaxivity of Gad complex chelate is equal in blood and myocardium (each assumed to be as single-compartments). If the relaxivities differ, r1 does not cancel out of partition coefficient calculation. This makes important the choice of GBCA selected and also may vary with 1.5T vs 3T.

Partition coefficient: Measure of relative distribution of Gad between myocardium and blood by the ratio of changes in their relaxation rates at steady-state of [Gd]. Can be two points (typically pre-Gad and 15ꞌ-20ꞌ post-Gad) or multipoint during washout (review, see Taylor et al 2016).

ECV corrects partition coefficient for variations in haematocrit.

Native T1 / postGad T1 / ECV : all have validations against histology of myocardial biopsies in diffuse fibrosis (Bull et al 2012; de Ravenstein et al 2015 ; Fontana et al 2012; Miller et al 2013; Iles et al 2015; among more).

For left ventricle, T1 is usually derived from set of single-shot images at different recovery times after inversion (MOLLI/ShMOLLI) or saturation (SASHA/AIR) or mixture (SAPPHIRE) (many after Messroghli et al 2004 and Higgins et al 2005; for refs see Roujol et al 2014; Taylor et al 2016)

The magnetisation recovery curve is pulled towards zero by measuring Mz (pitfall: often not shown on diagrams). This requires correction of the curve-fitted T1* to estimated T1 (Look and Locker 1970) depending on method. There are recent prospects of optimising this correction by sequence-simulation approaches. Even after correction, some T1 methods have systematic bias (e.g. possible influence of magnetization transfer Robson et al 2013)

Automatic online processing typically fits recovery curves to obtain a T1 value in each pixel of the images assuming constant cardiac and respiratory phase over all the recovery-times imaged. The output is known as the T1 map (do not use T1* map by mistake!) Limitations that users MUST check ALL of the input images to the mapping process, to guard against corrupted map data due to misgating or resp motion that may not be visually evident in the map.

Ensure T1 method has minimal heart-rate sensitivity.

Single-shot imaging → coarse image resolution, long shot duration, partial-volume error in myocardial pixels by adjoining bright tissues.

Imperfect inversion and saturation RF pulses eg short myocardial T2 (approx. 50ms).

Off-resonance and B1 transmit field calibration errors.

Regional errors over heart, especially in long native T1

(Chow et al 2015; Kellman P et al 2014x3; Kellman P et al 2013 x2).

Septal measurements of native T1 more reliable (ConSept, Rogers 2013).

Such systematic and random errors may be larger than small early fibrosis-related changes in T1.

Strongly confounding systematic differences occur between T1 methods, and even with nominally the same method after apparently small parameter changes (such as PE FOV impacting shot duration) and software updates. Each method requires normal-subject reference ranges, keeping the method frozen as far as possible thereafter, with regular QA scans to protect against unexpected changes in measurement (Captur et al 2016), distant prospect of worldwide standardization to a reference.

n.b. Some other myocardial applications of T1&ECV are clinically diagnostic, omitted as this talk is on diffuse fibrosis

Reference values have been obtained by large multicenter normal-subject studies (Dabir et al 2014; Rogers et al 2013; Piechnik et al 2013 and more)

Limitation - the values are protocol dependent and not necessarily transferable.

Many native T1 , postGd T1 and ECV clinical research studies show differences between normals and patient groups linked to diffuse fibrosis.

Limitation - findings are not specific to diffuse fibrosis. Interstitial space may increase for other reasons: inflammation (edema/myocarditis), other molecules than collagen may also “infiltrate” interstitial space. Intracellular space may decrease by death of myocytes (e.g. ageing). Genuinely increased diffuse fibrosis may be concealed from MR ECV (e.g. if accompanied by cardiomyocyte hypertrophy).

For diffuse fibrosis, T1&ECV scatter is too wide to enable strong clinical utility per patient (as opposed to clinical research groupwise statistical comparisons). Such a clinical test would be immensely valuable in cardiology as there is no non-invasive method for earlier detection of initial changes in diffuse fibrosis, which might be reversible without permanent impact if detected early enough.