The Mystery of Yttrium

There was some great science at SIR this year. I found the abstract session on radioembolization particularly thought-provoking. A series of fascinating presentations from Marnix Lam (catch his lecture at WCIO this New York this month!) examined dose distribution following resin-based radioembolization in metastatic disease. It was one of those “duh-oh” moments when he illustrated how body surface area based dose calculations result in undertreating little people with big livers and overdosing big people with little livers. There was a 5-fold range in liver dose from highest to lowest, with increased toxicity in the high-dose patients and lack of tumor response in the low-dose patients. 

The penetrating question from the audience (guess who?) was -- so what is the ideal liver-volume based dose of resin microspheres for metastatic disease?  Clearly the 120Gy target for glass microspheres is unattainable with resin microspheres. The median liver dose in Lam’s analysis was 50Gy, above which both response and toxicity were observed, but he was unable to prognosticate a recommended dose based on the limited retrospective analysis performed.

The mystery continued with presentation of Northwestern’s dose escalation study of glass Y90 microspheres with capecitabine for metastatic colon cancer, in which liver does of 170 Gy have been reached without increase in toxicity. So how can >50 Gy of resin Y90 be toxic, and 170 Gy of glass Y90 be safe?

Yttrium radioembolization sure is a mystery.


mystery is why it is a mystery....

May 14, 2013 01:31 AM by Dave Liu, MD

Thank you for bringing this 'controversy' to light....everyone take a deep breath because I would like to take a deep dive....

One question based on a few comments:

1) All Phase III data pertaining to the use of Y90 with resin microspheres is based on BSA method, including survival benefit.  BSA in these studies were based on clinical outcomes of undreds of patients not theory.


2) The currently recruiting prospective Phase III trials with resin Y90 (SARAH n=400, SIRVENIB n=280, FOXFIRE = 500, SIRFLOX n=500, FOXFIRE n=500, and SORAMIC
n=28) use BSA based formula


3) Kennedy et al recently presented data from MORE (mCRC Outcomes Radioembolizatio) in ASCO GI 2013 representing over 600 patients, demonstrating across the board encouraging results and acceptable toxicities with BSA method


4) The Asian and European communities are moving towards dose lowering not escalation (with a range of 40Gy to the liver, or 0.8GBq per liter). Extensively published consensus panel guidelines from both continents emphasize the need to lower activity and dose.


5) ‘Simplified’ partition calculations that are currently clinically used do not take into account how much actual radiation is going to the tumor and how much is going to the normal liver, which is contrary to many of the original studies that used much lower activity due to the fact that they took into consideration the uptake of the tumor vs the normal liver


6) Toxicities reported in the small retrospective cohort that has been described above did not account for increasing tumor burden and tumor lysis syndromes.  Furthermore toxicities were reported as CTC v3.0 Grade 1-4, not 3-4 (which is usually used to measure true toxicities).


7)  The discussion regarding underdosing is a circular argument, as the authors are using simplified partition model as a basis for this argument. Its like saying that: ‘I think the Toronto Maple Leafs are better than the Chicago Blackhawks. By measuring how much I like the Maple Leafs compared to the Blackhawks without comparing how they actually play, I conclude the Maple Leafs are the best’. (which we know is not true because the Vancouver Canucks are clearly superior!!!!)  Its not logical and its not clinically based.


It has to do with tumor coverage......

If you think of the tumor as a dark room with all sorts of nooks and crannies, you can light the room up more evenly with many low level lights (with less energy) distributed evening within the room as opposed to a single bright light in the center blinding in its center and casting long and uneven shadows/light in the corners.  In this way, BSA method using lower level radiation with a higher number of particles provides more even distribution with less overall radiation/energy.

The principle of crossfire (ie overlapping clouds of radiation) is a very well known radiation physics and radiation oncology but one that is rarely discussed in the IR community.  This principle, IMHO sheds light (or in this case beta particles) on the situation…..

Lighting up the dark room!

May 17, 2013 05:22 PM by Marnix Lam

Thanks for bringing up this interesting and important discussion. It clearly is controversial. 

If you would like to see a tumor as a dark room, the main question is: how big is the room!? A 200 Watt light bulb in the toilet is overkill, while a 20 Watt light bulb in the living room won't do! The spread of light is secondary.

The main point is that we need to know what we are doing when it comes to radioembolization dosimetry. With the activity planning methods used today (both resin and glass methods) that clearly is not the case. Dosimetry (i.e. what is the tumor dose, AND what is the normal liver tissue dose?) is totally lacking. The proposed partition method unfortunately is only feasible in a limited number of cases (one or few well demarcated tumors).

Chosing from 2 poor options, I choose the glass 'whole liver partition' method, because at least this method is based on some rationale (target size/weight). Nobody has ever been able to explain me the rationale behind the BSA-method?

The fact that we do so well without dosimetry makes me wonder how much better we can do with dosimetry...

Marnix Lam



a million points of light! and one dark spot...'dosimetry'

May 18, 2013 04:38 PM by Dave Liu

Dr Prof Lam,

Cannot agree more regarding the controversy and challenges.  Your comments regarding the differentiation between activity administration and dose/dosimetry is the Achilles heel of radioembolization and is certainly a deeper understanding of what is happening on an intratumoral level. To paraphrase your previous comments 'we got what we got and thats all we got!' (for the moment).
That being said, given your expertise in the field, would really appreciate your input and understanding on a few terms that are used and perhaps misused in casual conversation, for I think this will help the readers and myself better understand the perspective. Clearly this blog will not mark the end of the discussion for with a deeper understanding and iterative process we will only be one small step closer to the answer…
1) Could you define what you are referring to when using the term ‘PARTITION’? Is this the MIRD based compartmental model which looks at a specific volume and assumes uniform distribution?  If that is the case, when you are referring to dose, are you referring to an assumption of radiation distributed evenly within the tumor, evenly within the targeted volume (eg right lobe) or evenly within the liver parenchyma?

With your mention of ‘noone explaining how BSA modelling was determined’, as a counterpoint could you expand on how MIRD based partition was determined and more importantly how it accounts for degree of tumor involvement?  I have had heard some wonderful lectures on the mathematical modelling and evolution of the MIRD based model, (developed by Dr Mike Stabin) that have had him quoted as stating that the current state of the art MIRD based model represents bad dosimetry and that excellent dosimetry is still an area of research.

2) To my knowledge the BSA model is a regression analysis of the first few hundred patients undergoing SIRT procedure incorporated into these initial principles (and clinically validated in many prospective trials):

a. Normal liver volume is directly related to BSA (of course with any biological equation, the extremes are excluded from the distribution curve), which is a validated and accepted principle in FLR determination for portal vein embolization, as well as in transplant (Vauthey Formula, Vauthey et al Liv Trans 2002, Urata Hepatology 1995)
b. There has to be a compensation for the degree of liver enlargement from the degree of tumor infiltration, taking into account vascular capacitance of the tumor and uniform radioactive particle distribution (ie microdosimetry) if lobectomy or segmentectomy is not intended (ie 2 compartment model).
c. The liver parenchyma must be spared from excessive radiation to minimize the chances of Radioembolization induced liver disease (REILD) (REBOC consensus, IJROBP 2007, Kao JNM 2012, Gil-Alzugaray Hepatology 2013)

3) Adhering to the As low as reasonably acceptable (ALARA) principle of radiation exposure, and given the fact that the only limited data that is available is from very small and cross comparative studies, essentially demonstrating equivalent clinical and imaging based outcomes), what would the rationale be of administering more radiation that is necessary to elicit response? Treat/explant data dating back to 2001 (Fox et al IJROBP 1991, Kennedy et al IJROBP 2004, Campbell et al Phys Med Bio) have demonstrated intraoperative correlative exposure of tumors that are well above lethal range using BSA models.
No biological mathematical model fits in all situations, and oftentimes with standard therapeutic developmental pathways, determination of the Dose Limiting Toxicity (DLT) and Maximum Tolerated Dose (MTD) are key components in Phase I clinical trials designs to determine maximum threshold of therapy.  Unfortunately the evolution of both BSA and partition have not looked at this model before clinical acceptance and validation had occurred and as a result 20 years later we are still having the same debate.
I am sure that the readers of this blog will look very much forward to the continuation of this discussion, for I would suggest that the next segment should perhaps do with EX partition modelling, microdistribution, monte carlo based dosimetric modelling and crossfire/electron track prediction which should be fun!


Dank u!
(a diehard Canucks fan…they just had a bad year)

We can do better!

May 20, 2013 11:46 AM by Marnix Lam

Dear Dave,

Thanks for your interesting reply. The quest for further improvement of this very powerful treatment modality is feeded by the experience of many of us, that many patients respond poorly, but also because some patients experience unexpected toxicity. First of all, we need to know why? Besides obvious patient related factors, we also need to look at the treatment itself, and its effect. Because currently, radioembolization is a shot in the dark. As said, proper dosimetry is largely lacking and dose-response relationships do not exist. With optimized dosimetry and treatment planning I believe we can go towards complete responses in much more patients with acceptable and controlled toxicity in all. This will proof especially important in combined treatment regimens, which is anticipated in the near future.

To answer your questions:

1) I would like to reserve the meaning of the word 'partition' for segmentation of the liver in different compartments. Of cours this is ultimately needed in every dosimetry approach to differentiate between tumor and normal liver absorbed dose. Within a selected compartment one may use the 'MIRD' based assumption of homogeous distribution (Gulec et al. JNM), which is obviously not the case (and like you said 'bad dosimetry'), or better, one may (and ultimately should) use a voxel-based approach and dose-point kernels to calculate not only the mean absorbed dose within the compartment, but also the spread of the absorbed dose within that compartment using voxel-based dose-volume histograms in analogy with radiotherapy practice. With other words: what percentage of the tumor received a dose > 70 Gray?

2) The scientific basis for the BSA-method has never been published. Since it calculates activity it should be directly related to some sort of dosimetry parameter to have some sort of rationale at all, even a poor parameter, anything at all. I challenge anybody to show us this!? 

3) As soon as we understand liver dosimetry better we should perform a dose-escalation study based on this understanding with the absorbed dose to the normal liver tissue as a dose-limiting factor.

Why more activity? Because, we CAN do better. Understand treatment principles, create expected dose-response raltionships and treat accordingly. More (complete) responders, even less (and understood) toxicity! In selected patients more acticity will lead to a higher absorbed tumor dose and better response!





May 20, 2013 09:12 PM by Dan Sze

1.  We have not proposed escalating the dose.  We are searching to find a better method to customize dosing to a patient’s disease.  In fact, we hope to identify prospectively the 1-2% of patients who are currently destined to suffer from REILD, who should be given smaller doses or not treated at all.  Also, we hope to identify the 20-70% of patients who do not respond, and within that group identify those that could tolerate and benefit from a higher dose.

2.  In a large room, one 500 watt light bulb is probably less desirable than fifty 10-watt bulbs.  How many is enough?  Are 500,000 one milliwatt bulbs even better?

Assume volume of distribution is 2,000,000 cubic millimeters (about the average liver, 2 liters).  If we give 8 million microspheres (approx 20 GBq glass dose or 0.4 GBq resin dose) results in an average 4 microspheres per mm3.  With a maximum tissue penetration of 11 mm for the beta particles, each point sees radiation from a surrounding volume of 5572 mm3, or about 22,290 microspheres.  With average tissue penetration of the beta particles of 2.4 mm, each point sees highest radiation from a surrounding volume of about 58 mm3, or about 232 microspheres.  The hypervascularity of tumors should result in higher numbers than this in the targeted volume.  Is this enough crossfire?

3.  What if there are 2 rooms, one you want to light up, one you want to keep dark.  If the room to be lit is only big enough to fit 50 bulbs, but all you have are 5 watt bulbs and you want to have a total of 500 w, you need to try to cram in 100 bulbs.  After 50 you reach capacity (stasis) and any additional bulbs will spill over into the room you want to keep dark.

4.  The Sharks (from San Jose, where “ice” is what you blend into your margarita and “skating” requires a skateboard) defeated the Vancouver Canucks 3-0 in the playoffs.  The statistics are not circular logic.  Even though you rooted for the Canucks, the score objectively favored the Sharks.  Even though we rooted for consistent 100% efficacy and 0% complications using the current dosing recommendations, the statistics show otherwise.  This is not a flaw of the resin product, which has shown good results and we continue to use and to endorse.  While “better” may be the enemy of “good,” “good enough” is the enemy of “optimal.”  Clearly, there is room for improvement, and this dialogue is the way to get there.

My brain is starting to hurt...must be too many crossfiring neurons....

May 25, 2013 04:02 PM by Dave Liu

I.  BSA or BS?  MIRD or MUD? 

I think we are all speaking of the same point regarding the deficiencies of both systems. However that being said, inherent to the BSA methodology is the incorporation of a corrective factor for degree of tumor infiltration that is reflective of biological systems and based out of clinical data. As much of a black box as it may be, the equation:

BSA Activity = (BSA-0.2)+TI

Is a baseline, or starting point of minimization of toxicity.


To put it another way:


Is the starting point to an ALARA based (minimum required) amount of radiation,



TI = (tumor volume/total volume) 

Is the corrective factor of increasing radiation based on whether there is more or less tumor.


As with any biological system, once you go two or three times beyond the standard deviation exceptions occur (for instance, look at the Theory of Relativity, it breaks down as we approach the speed of light!). Admittedly personally I do make adjustments both up and down based on my perceived concerns of parenchymal exposure (eg low tumor volume, cirrhosis, chemotherapy associated liver changes) or increase (not based on ‘giving more dose’ but based on tumor volume coverage; a subtle but important difference).  As I have heard and taken the lesson to heart from the pioneers in the field such as Andy Kennedy and Riad Salem ‘safety before efficacy’.


Contrary the ‘baseline safe activity + additional activity for additional tumor’, single compartment partition modeling (the current commercially recommended method) is inherently flawed as it does not take into account any aspect of target tumor volume.  For those that are still reading this blog, there is a very good thumbnail rule of 1-1-50 for glass partition modeling….1kg of liver given 1GBq of Y90 activity will have an absorbed ‘dose’ of 50Gy….

Here are three scenarios, all starting with the basic premise….let us target 1kg of tissue, let us administer 1GBq of activity to this 1kg of tissue in the following scenarios

  1. 1) 1kg normal liver

  2. 2) 500g tumor, 500 g liver

  3. 3) 1kg of tumor

According to the commercial single compartment method, all three of these situations are the same….

waaaaiiiit a minute….what about hypervasularity, what about tumor uptake and what about the actual liver parenchyma?  How does a dose determination using such a crude method many any sense? 


I am confused because my interpretation of this situation is that the parenchyma is receiving a variable exposure depending on how much tumor there is with no capacity to control this variable….that doesn’t make any sense either...


II.  Illumination or Illuminati? 

Regarding the mathematical principles described by Dr Prof Sze, I agree with particulate distribution however compartmentalization of the particles is the foundation of therapy.  Given that the microdosimetry (ie how the particles distribute within the tumor, which then translates into how much exposure of radiation there is, which translates into response) is to a large degree not factored into either BSA or partition, the discussion is mostly academic and somewhat of a clinically moot point (hopefully with more and improved modeling it will factor into clinical domains but for now, it hasn’t).

So, going back to the comments and analogies of the lights in a dark room, the objective is to cast as much even light in the room (ie tumor) without blindingly bright point sources, and without casting long and harsh shadows, to this I think we can all agree.  The point I was trying to make in that analogy is that overall watts (actually lumens…to be accurate) does not translate into overall improvement in the lighting of the room, it is the distribution of the lights in the room that provides even and uniform light.

This addresses the issue of crossfire; if we look at low level light sources that work together as opposed to a single bright light bulb, the quality of the light (ie radiation) is much more even (ie crossfire).  

Furthermore this issue is just as important to the tumor as parenchyma…what do you think is happening when we see edema in a treated lobe, or geographic areas of fissure-like edema post administration, or peribiliary edema…it is the microclustering of high activity particles in normal anatomy that is being damaged….using lower activity particles decreases the cumulative effects of radiation in normal anatomy (at least that is what I observe with resin and glass EX methods of administration)


III. The Mystery of Yttrium: Remix

To this point, perhaps going full circle back to Dr Prof Soulens comments, it is a mystery as to what the simple elegant solution is, for with the tools we currently have we are trying to measure the diameter of an atom with a meter stick regardless of what type of window dressing we are putting around our opinions….I JUST HEAR FROM TOO MANY PEOPLE THAT ‘MORE ACTIVITY NEEDS TO BE ADMINISTERED’ RATHER THAN ‘BETTER DISTRIBUTION OF ACTIVITY NEEDS TO BE PLANNED’ and without an understanding of 3D crossfire, isodose cloud distribution, microclustering and microdosimetry, the discussion cannot be effective or productive as it will reduce down to an argument of  ‘I like this better because I do…’


It is as complicated as you make it.

May 30, 2013 01:52 AM by Bulent Arslan, M.D.

Why do we want 2000 watts? If the goal is to see better, the brighter the room the better you will see. If that much light is going to blind somebody, maybe then just make as bright as possible without overkilling. Enough radiation will kill anything in its territory. With high enough dose, there will not be a question of efficacy, but obviously there will be safety concerns. When you administer over a 1000Gy to a segment that is no different from surgical segmentectomy, except it does not have the complications and recovery issues of surgery. The poroblem is than not all tumors are amenable for that approach and if they are, they end up getting resection or transplant.

I believe Resin and glass based Y90 is very different from each other. Resin based y90 is more comparable to  40 micron bland beads than the glass based y90. Its embolic effect likely has more treatment effect than the radiation it provides. That's why it probably is approved for colorectal and glass based y90 for HCC. It is definitely anectol but in our experience resin performs poorly in HCC and we had to retreat quite a few patients with glass to eventually get a better response.

Figuring out dosimetry is important, but let's face it by lowering the existing dose incresing efficacy is quite unlikely. If there is room to improve efficacy, beyond more targeted delivery it will be by genetic profiling of the tumor to see if they are more sensitive to radiation and mainly by increasing the radiation dose as long as it is within safety limits.

With resin based I doubt there is much more room to increase the radiation dose as you will not be able to achieve it in most cases due to reflux, but with glass based there is that potential. Additionally resin based beads are not only less radiactive but also slightly larger in average diameter (5 micron?). That does not help either.

Bulent Arslan

Dosimetery my good man - it's all dosimetry!

May 31, 2013 11:45 PM by Fred Martin Moeslein, MD, PhD

Michael, as usual you’ve initiated a creative thought experiment, which has drawn together a decidedly lively and exceptionally thought-provoking discourse. After all of this wonderful banter, I too feel compelled to dive into this line of thinking.


First, I believe the West of the Mississippi crowd has completely missed the boat here. The truly superior team is obviously the Pens, and it goes well beyond the exceptional play of the many deep lines on this certain to be Stanley Cup championship team, and originates from their name sake. As all learned observers would clearly acknowledge, ONLY PENGUINS BELONG ON THE ICE! QED


As for the lesser issue of BSA vs. MIRD vs. your dose estimator du jour, they are all dose estimations (please insert any synonym you prefer but its all a guess). Rather than rail against one guess or another we should ALL be asking each other how it is that NONE of us has any clue how much radiation we actually need, nor how to get it where we most need it. No offense to the brilliant forefathers who have led us to this wonderfully effective therapeutic option, but we have arrived here devoid of any true predictor of needed dose. In the absence of science showing us how much dose any particular cancer, in any particular patient, is actually needed to kill said cancer, we should all be attempting to minimize toxicity first and foremost. (And then running back to our labs to design experiments to determine calculations that will allow us to maximize efficacy in the future.) FIRST, DO NO HARM.


Yes we can all point to the anecdotal mNET patient with a huge liver and a tiny body that gets “under-dosed” – but HOW DO YOU KNOW THEY WERE UNDER-DOSED?? As far as I can tell none of us actually has a gold standard to compare tumor kill between treatment modalities (and by the way that holds true for TACE and Bland embo too). If I am wrong please correct me, but what is our gold standard? How do we know that this “under-dosed” tumor that didn’t die wasn’t in fact actually a radiation insensitive clonal variant that was in fact “over-dosed” but just wouldn’t die? The fact is Dr. Lam your assertion is based on a perception of under-dosing because there is “less radiation” not Proof of under-dosing. Dr. Sze you correctly point out that we need to determine what exactly makes tumors more or less susceptible to our Y90 therapies. That may be the need for more dose or more particles (or more dose on more particles) – or maybe its less dose delivered repeatedly to attack a greater number of at risk tumor populations – or maybe its way less dose and way more induction chemotherapy – or maybe its none of those and the true answer lies down a completely different avenue. As we however, lack a gold standard to measure these assertions against, a tacit statement such as “under-dosing” is essentially unsupportable on scientific grounds. It may be correct, but we cannot prove it, because NO ONE knows what “adequate dosing” is beyond the generic level of “well the tumors didn’t all die so we must have under-dosed” or the “liver was overly damaged so we must have over-dosed”. Those are pretty weak arguments, which have been wrapped up in some pretty heady suppositions, but in the end without knowing what dose you actually need how can you actually pretend to know if you got it?


Furthermore, not only do we lack an understanding of how to predict necessary dose we cannot actually (except in animals and potentially in rare experimental settings) even say for sure how much dose a given tumor or segment of liver actually received DURING a Y90 treatment. That is not a glass thing or a resin thing that is a “scientific deficit” thing – we LACK DOSIMETRY. How are we supposed to reach an “accurate prediction” of required dose when we can’t even decide on the DOSE DELIVERED to any given tumor or segment of liver!?!


I would second Dr. Liu’s point: All biological models fail at the extremes, as they are inherently based on assumptions, and as we learned in grade (or for you late bloomers – middle) school – we all know what happens when you ASS-U-ME! I would welcome a dialogue about how we can better resolve some of these weaknesses in these systems, but I would argue that the experienced user colors their treatment calculations with a lot more nuance than this discussion of BSA or MIRD would initially engender. Who here actually treats a fourth line mCRC salvage patient with even the full-dose BSA calculated dose? What about the ECOG 2 patient with HCC? Or the PS 2 breast cancer patient who is now 8 years into her chemotherapy treatments? AND WHY NOT? Because none of us are actually sure what our radiation is going to do to those patients. AND WHY IS THAT? Because we don’t actually know how much radiation we need to kill those tumors nor how much damage the errant particles will cause beyond the tumor margins. Again, we need accurate post-treatment dosimetry on hundreds or thousands of patients so that we can begin to develop truly prospective models to improve efficacy, not just minimize toxicity.


Is the BSA flawed? Certainly. But so is the Mona Lisa – which is one reason we all stare so intently at her visage. I would suggest the real focus here shouldn’t be the current estimation model that we use for glass or resin dosing, it should in fact be maintaining a safe treatment paradigm for the majority of our patients until such time as we can better predict necessary dosing requirements for all of our patients. And I would add, these requirements go beyond tumor vascularity, tumor volume, and liver volume, as they extend to tumor type, prior chemotherapy, ongoing chemotherapy, extent of clonal variance and genetic sensitivities, just to name a few. The BSA model for all of its flaws has been tested in hundreds (if not thousands) of published patient experiences, and shown to be safe and for the most part effective. If the partition model were so much better why is it that we are not curing all with its numbers? Because it is flawed too! Someone wiser than me reminded me (with regard to Y90) that you can always add more at a later time, but you can’t take back what you’ve already put in. PRIMUM NON NOCERE!


And I am spent!








yttrium radioembolization

June 2, 2013 08:44 AM by Mark Westcott

This discussion has been informative and thought provoking.  The
challenges we face delivering a tumoricidal dose while limiting normal liver
dose seem numerous but they are not insurmountable.  While the methods we currently employ have many weaknesses literature to date confirms therapeutic efficacy with both
glass and resin as discussed by others above. But of course we can do better.  I agree with Dr. Liu that more delivered activity is not the answer and that this push by some that giving more will lead to better outcomes is very likely to instead lead to increased morbidity and
mortality.  The goal should be to give the correct dose to each patient using dosimetric techniques based on imagingand T/N ratios. Improved dosimetry methods have been recently described by various investigators including Kao et al from Singapore (CT Hepatic
Arteriography and SPECT/CT: Obtaining More Accurate Partition Modeling, Jnl Nuc
Med Feb 2012.).

 An interesting read is Recommendations of the American Association of Physicists in Medicine on Dosimetry, Imaging and Quality Assurance Procedure for 90Y microsphere
Brachytherapy in the Treatment of Hepatic Malignancies. WA Dezarn 2011. (take
it to the beach).  This document outlines the BSA and partition models currently in use and the disadvantages of both.  Of interest is a point often made but sometimes forgotten that the partition model “does not take into account the activity non-uniformity within each partition” and it “cannot be accurately used for diffuse tumor where extent cannot be determined”.  One of the final recommendations by the panel for further research is to “Replace the MIRD based, BSA or other empirically prescribed activity methods with careful 3D dose distribution and dose volume histogram treatment planning”.   One final point about current dosimetry methods is that while resin users typically employ the BSA method they may also use the partition model as outlined in the Sirtex Medical Training Manual pages 34-37. 


Dr. Arslan, I would like to address your comment that the effect of resin microspheres is attributable to their embolic nature, similar to 40 micron bland embolization.  The published and clinical evidence to date unequivocally refutes this contention.  Bilbao and his
group have clearly shown that resin microspheres have no embolic effect on
median to small arteries.   While reduced flow or stasis is seen during
resin microsphere administration, it has not been determined to be caused by
the microspheres.  Many resin users now feel that reduced flow (sometimes seen quite early in the administration) may in large part be secondary to vasoconstriction in response to the injected sterile water.  The necessity of sterile water is being investigated.  Additionally it may
be that reduced flow/stasis is an oft cited but possibly insignificant issue in regards to ability to deliver the prescribed dose.  Kennedy et al showed that in a cohort of 680
patients treated with resin microspheres the administered dose was only
slightly smaller (5-10%) than the prescribed dose. ( Kennedy AS, McNeillie
P, Dezarn WA, Nutting C, Sangro B, et al. (2009) Treatment parameters
and outcome in 680 treatments of internal radiation with resin 90Y-microspheres
for unresectable hepatic tumors.
Int J Radiat Oncol Biol Phys 74:1494-1500.)  This small amount can be attributed to residual spheres within the tubing and V-vial at the end of
delivery (which is why my group draws 10% more than prescribed) and is seen
even in cases where there is no reduced flow. 

Additionally you report poor response in your HCC population following resin microsphere administration.  That has not been my experience. There are quite a few publications that show similar overall survival for glass and resin microspheres:

Dancy JE et al (2000) Treatment of nonresectable hepatocellular carcinoma with intrahepatic
90Y microspheres. J Nucl Med 41:1673-1681


Salem R et al (2010) Radioembolization for hepatocellular carcinoma using Yttrium-90
microspheres: a comprehensive report of long-term outcomes. Gastroenterology


Sangro B et al(2006) Radioembolization using 90Y Resin microspheres for patients with
advanced hepatocellular carcinoma. Int J Radiat Oncol Biol Phys 66:792-800


Lau W et al 1994 Treatment of hepatocellular carcinoma with intra-hepatic arterial
yttrium-90 microspheres: a phase I and II study. Br J Cancer 70:994-999


Sangro et al 2011 European network on radioembolization with 90Y resin microspheres
(ENRY) multicenter evaluation of survival, safety and liver function for
unresectable hepatocellular carcinoma(HCC). J Hepatolo 54:S36

how do we know what we are giving?

June 3, 2013 10:33 PM by Alain Drooz

Thanks for all the elegant arguments.  Take it down to a more basic level for a moment, and I think all of us will agree that both the partition and BSA methods are approximations.  We have no way of determining the exact distribution of the microspheres, and we all have to make adjustments for activity based on patient factors.  What do you want, quantitative analyses of tumor/normal ratio integrated over the entire volume of tumors throughout the liver??  

There is something different about the two products.  If it takes 120 Gy with glass to get the same therapeutic benefit as 50 Gy with resin, it doesn't seem to me that the problem is with the resin...


Alain Drooz

Are we building a foundation of science or tower of Babel?

June 5, 2013 01:23 AM by Dave Liu

Dr. Arslan,

Your comments are certainly from a strong opinion and personal experience but I would like to reference the published literature as there are several factual errors in your comments:

I.  Less Filling or Taste Great?
 Your Comment: ‘I believe Resin and glass based Y90 is very different from each other. Resin based y90 is more comparable to 40 micron bland beads than the glass based y90. Its embolic effect likely has more treatment effect than the radiation it provides. That's why it probably is approved for colorectal and glass based y90 for HCC. It is definitely anecdotal but in our experience resin performs poorly in HCC and we had to retreat quite a few patients with glass to eventually get a better response.’

 May be a reflection of your personal experience of which I would defer to your personal opinion and observation on but from both a clinical and scientific side, European thought leaders, including Dr. Prof Sangro and Dr. Prof Bilbo have published extensively on the following:

 Emboli effect: (this won the Abstract of the year at CIRSE)

 A)    Bilbao, J. I., de Martino, A., de Luis, E., Díaz-Dorronsoro, L., Alonso-Burgos, A., de la Cuesta, A. M., ... & de Jalón, J. A. G. (2009). Biocompatibility, inflammatory response, and recanalization characteristics of nonradioactive resin microspheres: histological findings. Cardiovascular and interventional radiology, 32(4), 727-736.

 ‘Eight-week evaluation found that the perivascular inflammatory reaction was mild. Liver cell damage, bile duct injury, and portal space fibrosis were not observed. In conclusion, resin microspheres (15–30 μm diameter) trigger virtually no inflammatory response in target tissues (liver and kidney). Clusters rather than individual microspheres were associated with a mild to moderate perivascular inflammatory reaction. There was no evidence of either a prolonged inflammatory reaction or fibrosis in the liver parenchyma following recanalization.’

 Translation: there is no terminally embolic effect associated with resin microspheres; there is essentially no response effect as a result of the physical properties. As I am sure you are aware, this is fundamentally different than DEB and TACE.

B) Simon MacKie, Suresh de Silva, Peter Aslan, Leigh Ladd, Michelle Houang, David Cade, Warwick Delprado. Super Selective Radio Embolization of the Porcine Kidney With 90Yttrium Resin Microspheres: A Feasibility, Safety and Dose Ranging Study, J Urol 2011 (volume 185 issue 1 Pages 285-290 DOI: 10.1016/j.juro.2010.09.001)

 ‘We performed super selective radio embolization with 90Y resin microspheres in 1 kidney and with an equivalent number of bland microspheres in the corresponding pole of the contralateral kidney as a control.’

Results: No discernable effects of the non radioactive microspheres in the normal porcine model, with effects of damage related directly to the amount of radiation used.

Translation: particles are too small to cause an apoptotic or hypoxic inflammatory reaction.


 C)    Clinically, Resin for the treatment of HCC in the setting of PVT have been published refuting your comments regarding the primary embolic effect of resin microspheres for as we all know, PVT is the exact scenario in which embolic is less desirable.

Articles from Pamplona with resin and PVT:

 Inarrairaegui, M., K. G. Thurston, et al. (2010). "Radioembolization with use of yttrium-90 resin microspheres in patients with hepatocellular carcinoma and portal vein thrombosis." Journal of vascular and interventional radiology: JVIR 21(8): 1205-1212.

 ‘Radioembolization of unresectable HCC and branch or main PVT with (90)Y resin microspheres was associated with minimal toxicity and a favorable median survival time. Further prospective studies are warranted to validate the findings in this clinically challenging patient population.’

i. Article from Pamplona with resin and PVT:

Lau, W. Y., B. Sangro, et al. (2013). "Treatment for hepatocellular carcinoma with portal vein tumor thrombosis: the emerging role for radioembolization using yttrium-90." Oncology 84(5): 311-318.

 ‘Transarterial radioembolization (TARE) with yttrium-90 microspheres is emerging as a valuable strategy. A wider range of patients with PVTT are suitable for this procedure compared to TACE. TARE is as effective as TACE in HCC and has quality-of-life advantages. Conclusion: In patients with HCC with PVTT, medical evidence suggests that TARE is a good choice of treatment.’

ii. Article from Singapore with resin and PVT:

Burgmans, M. C., F. G. Irani, et al. (2012). "Radioembolization after portal vein embolization in a patient with multifocal hepatocellular carcinoma." Cardiovascular and interventional radiology 35(6): 1519-1523.

 ‘Radioembolization is an effective locoregional therapy for patients with intermediate or advanced stage hepatocellular carcinoma (HCC). It has been shown that radioembolization is safe in patients with portal vein thrombosis... describes safe radioembolization after portal vein embolization in a patient with multifocal HCC.’

II. Will More Radiation Cook a Tumor ‘Dead’er’?

Your comment: ‘Figuring out dosimetry is important, but let's face it by lowering the existing dose increasing efficacy is quite unlikely. If there is room to improve efficacy, beyond more targeted delivery it will be by genetic profiling of the tumor to see if they are more sensitive to radiation and mainly by increasing the radiation dose as long as it is within safety limits.’

The comments are interesting as I would like to know what your thoughts are on the common practice of using Glass EX ‘dosimetry’ and lowering the partition model ‘dose’ to 80Gy. Many high volume and experienced institutions do this on a regular basis in order to improve tumor coverage.  I would assume that would mean that you do not believe or do EX treatments?

III. How Many Fairies Fit on the Head of a Pin?

Your comment: With resin based I doubt there is much more room to increase the radiation dose as you will not be able to achieve it in most cases due to reflux, but with glass based there is that potential. Additionally resin based beads are not only less radioactive but also slightly larger in average diameter (5 micron?). That does not help either.

Dating back to 1957 with the seminal paper by Breedis and Young, with the initial descriptions of hepatic tumor vascularization and subsequent work by Ackerman and Lein in the 1970’s the hypervascular rim that constitutes the target of carrier based deliver mechanisms are in the range (I emphasize range) of 30-70um.  5um is 1/10 the width of a human hair, half the size of a red blood cell and closer to the size of a bacterium…

Targeted embolic therapy (bland) has been effective in the 100-300 and sub 100 um range; that data has been very well published and validated by the Memorial Sloan Kettering group over the last 15 years.

Proximal embolization to increase AUC and diffusion/dwell time for lipiodol tace is in the 300-500um range, demonstrated by the Hopkins group a few years ago.

Anything bigger than 500um can result in collateralization and reconstitution of flow. This has been demonstrated in gelfoam  embolization dog studies in the 1980’s

In short the scale you are referring to of 5um is irrelevant based on published literature (and as mentioned is the size of a single bacterium), current knowledge of angiogenesis and physical property when it comes to tumor vasculature.

Regarding increasing the radiation dose, please define what you mean by dose? If you mean activity (i.e. GBq) that is the whole point of the discussion and would encourage a review of the previous postings and intellectually rewarding discussions with Dr. Prof Lam on the exact topic….

IV. IN SUMMARY, to which I wish to reiterate

I JUST HEAR FROM TOO MANY PEOPLE THAT ‘MORE ACTIVITY NEEDS TO BE ADMINISTERED’ RATHER THAN ‘BETTER DISTRIBUTION OF ACTIVITY NEEDS TO BE PLANNED’ and without an understanding of 3D crossfire, isodose cloud distribution, microclustering and microdosimetry, the discussion cannot be effective or productive as it will reduce down to an argument of  ‘I like this better because I do…’

Recent Stories
A Message From San Francisco Mayor London N. Breed
SIO2022 Postponement

A Note on Health and Safety at SIO2022