Latest / Pediatrics Now: Cases Updates and Discussions for the Busy Pediatric Practitioner / Silent Crisis: Pediatric Cancer Gaps in Latino Communities

Transcript

Silent Crisis: Pediatric Cancer Gaps in Latino Communities · Pediatrics Now: Cases Updates and Discussions for the Busy Pediatric Practitioner

0:01
I'm Holly Wehment, and this is Pediatrics Now, brought to you by the Department
0:06
of Pediatrics at UT Health San Antonio and University Health's new Women's and Children's Hospital.
0:13
Today, I'm bringing you two great talks from the Advancing Cancer Research for
0:19
Latinos at All Populations Conference.
0:22
This is from two doctors.
0:24
Why children are falling through the cracks of pediatric cancer care in America. Welcome in.
0:30
I hope that you all find this very valuable, exciting, and somewhat,
0:35
you know, something that can project in the future some benefit,
0:38
especially if we can coordinate more efforts together. That would be amazing.
0:43
So this panel, as I mentioned, is pediatric cancers, and we're going to call
0:47
it, you know, the silent crisis, why children are falling through the cracks
0:51
of the pediatric cancer care in America, especially, you know, the most vulnerable,
0:56
underserved patients. So our kids.
1:00
So we're going to start with Dr. Smith. Dr.
1:03
Smith is an associate professor in the Department of Pediatrics at the Beckman
1:07
Research Institute of City of Hope Medical Center and a scholar of blood cancer,
1:13
United, formerly the Leukemia and Lymphoma Society.
1:16
Dr. Smith is a genetic epidemiologist with research focused on identifying the
1:21
causes of acute lymphoblastic leukemia, the most common childhood cancer,
1:26
particularly in Hispanics.
1:27
His particular interest in undergoing increased risk of ALL in children of Hispanic
1:33
and Latino ethnicity would focus on the role of genetic variations and genetic ancestry.
1:38
So please, applause and welcome Dr. Smith. Thank you so much.
1:42
Buenos dias. Good morning, everyone. Very happy to be here. Thank the organizers
1:47
for inviting me to speak.
1:49
Actually, there's nowhere to put my water. So I'm excited today to speak about
1:54
our research on acute lymphoblastic leukemia,
1:57
which some of you may or may not know has an increased risk in individuals of
2:04
Hispanic or Latino ethnicity.
2:06
So first of all, I'll just give a brief background on what we know about the
2:11
epidemiology of acute lymphoblastic leukemia, and then some of our research
2:15
looking into the genetic contribution to the increased risk of ALL in Hispanic or Latino children.
2:21
And then finally, I'll briefly talk about some research from other groups that
2:26
have looked at genetic ancestry and somatic alterations and patient outcomes in childhood ALL.
2:32
So ALL is the most common childhood cancer. It's a cancer of white blood cells,
2:37
mostly B cells, about 80% and about 20% T cell ALL.
2:43
So it comprises about 30% of childhood cancers, and there are over 3,000 new cases each year.
2:49
Now, the good news is, if you look at this figure on the right-hand side,
2:54
cure rates of childhood ALL have really increased over the decades.
2:58
So it's really one of the success stories of cancer treatment.
3:03
And this was due to the development of combination chemotherapy that's led to
3:09
cure rates of upwards of 90%.
3:11
Now, the bad news is those survivors of childhood ALL face lifelong treatment-related
3:17
morbidities and increased risk of mortality, for example, from secondary cancers.
3:22
Also the bad news, ALL remains a leading cause of childhood mortality,
3:27
despite the high cure rates.
3:30
So really, prevention remains an important goal. In terms of the natural history
3:34
of childhood ALL, we know that for many of the molecular subtypes of ALL,
3:40
the first hit mutation actually occurs during pregnancy in utero.
3:44
And this has been shown from studies, for example, of identical twins,
3:49
where one twin has developed leukemia, and that same mutation has been found
3:54
in their identical twin who did not develop leukemia.
3:57
And it's also been shown from what's called backtracking studies,
4:00
where we can define the mutation in the tumor at diagnosis and then go back
4:05
to blood samples from birth, for example, newborn dry blood spots,
4:09
and find that exact same mutation.
4:12
We also know that for some of the molecular drivers of ALL, these can be found
4:17
in kids who never develop leukemia.
4:19
So for example, the ETV6-Rungs1 fusion, which can be found in about 25% of childhood ALL.
4:27
This fusion has been found in up to 5% of healthy newborns.
4:31
So we know that this initiating lesion can occur prenatally,
4:35
and this pre-leukemic clone can go on for several years and may eventually become
4:41
leukemia when their second or third hip mutations occur later in childhood.
4:46
So our group is interested in what are the genetic modifiers and what may be
4:51
the environmental modifiers, both of this initiating lesion and of these second hip mutations.
4:56
And for those of us who are interested in what causes ALL, we're starting to
5:00
think about the potential for newborn screening of pre-leukemia for earlier
5:04
detection and perhaps even prevention.
5:07
And I'll come back to this point at the end of my talk.
5:10
One of the most striking trends in the epidemiology of ALL is this increased
5:15
risk in individuals of self-reported Hispanic or Latino ethnicity.
5:19
And so this is data from SEER. You can clearly see the increased risk in Latinos.
5:25
And this is not just in children. This is actually across the age groups.
5:29
So in fact, when you get to the AYA and adult population, it's about a twofold
5:33
increased risk compared to non-Hispanic or non-Latino whites.
5:37
Interestingly, this increased risk appears to be specific to the B-cell immunophenotype
5:43
of ALL, which points to there being an underlying biological cause.
5:47
In addition to the increased risk, we know that the incidence of ALL has been
5:52
rising in Hispanic or Latino populations, and this increasing incidence appears
5:57
to be specific to this population.
6:00
Again, this is also seen across age groups.
6:03
And there are also racial and ethnic disparities in the outcomes of childhood ALL patients.
6:08
So both Hispanic Latino and Black or African American patients have inferior
6:13
survival than non-Hispanic whites.
6:16
This has been shown kind of 20 years ago looking at SEER data,
6:20
but has been shown to persist to recent years.
6:23
And there are likely several reasons for this disparity, including social determinants
6:28
of health, environmental factors, tumor biology, and genetics.
6:32
Including genetic ancestry which I'll talk about a little bit later.
6:36
And there's also evidence that Hispanic or Latino children with higher proportions
6:40
of indigenous American ancestry also have worse survival.
6:44
So one of our main research interests is in addressing the question of what
6:49
is the genetic contribution to the increased risk of ALL in Hispanic and Latino children.
6:54
And so I'll describe some of our recent research that tries to address this
6:57
question looking at both the contribution of genetic variants as well as genetic ancestry.
7:04
So Hispanic and Latinos are a genetically admixt population,
7:07
comprising the genetic ancestry proportions of each patient can be inferred
7:13
from their genotyping data,
7:15
where each individual's ancestry proportions reflects the fraction of their
7:19
polymorphisms across the genome that have frequencies most similar to those
7:24
in reference populations,
7:26
for example, African, European, and Indigenous American individuals.
7:30
And altogether, these ancestry proportions sum to 100%.
7:34
So we know that genetic ancestry is correlated with race or ethnicity,
7:38
but it captures more complex population origins and also can capture different
7:45
exposures and factors that may impact disease risk and patient outcomes.
7:49
And by studying genetic ancestry and the variants that are associated with ancestry,
7:54
this may provide clues as to the increased ALL risk in Hispanic and Latinos.
7:58
I did want to point out that there's four Hispanic and Latino populations in
8:04
this figure, and I'm not sure if my mouse works, but I'll try and describe it.
8:10
You can clearly see, for example, in the Peruvian population have much higher
8:18
proportions of indigenous American ancestry compared to the Puerto Rican population
8:23
that has a larger contribution of European and African ancestry.
8:28
In this review paper from a couple of years ago, we showed that many of the
8:32
known childhood ALL risk allele frequencies are skewed towards having a higher
8:36
frequency in Hispanic and Latinos than in European populations,
8:40
highlighting a couple of genes of interest here.
8:44
IKZF1, which I'll come back to in a couple of slides.
8:47
GATA3 and ARID5b have been known for several years now to have a higher risk
8:52
allel frequency in Hispanic Latino populations.
8:55
The ERG-SNP, although the risk allel frequency difference is not huge,
9:01
it actually has a larger effect on ALL risk in Hispanic and Latino children
9:05
than in children of European ancestry.
9:08
So we think this highlights the importance of studying genetic risk of ALL in
9:13
Hispanic and Latino populations.
9:15
Now, as well as looking at these individually, we can also sum the effects of
9:19
all of these individual variants into what is called a polygenic risk score.
9:23
And this is from a multi-ancestry study, genome-wide association study,
9:28
using our California-based ALL cases and controls.
9:32
We constructed this polygenic risk score from, at the time, 18 of these SNPs or variants.
9:38
And you can clearly see that this PRS is shifted to the right-hand side in Hispanic and Latinos.
9:45
So this was with cases and controls combined. And when we stratify by case control
9:50
status, you can see that even amongst the controls,
9:53
the Hispanic-Latina controls have a significantly higher polygenic risk score
9:59
for ALL than the European Ancestry controls.
10:02
So this higher burden of the ALL SNPs may partly explain the increased ALL risk,
10:09
although this may be simplifying things in terms of just comparing PRS across populations.
10:15
But beyond the genetic loci that were known, are there additional genetic factors
10:21
that contribute to the ALL risk in Hispanic and Latinos?
10:24
So previous studies, including our own, had identified IKZF1 as a risk locus for a childhood ALL.
10:31
But the number of the common alleles at this locus, the causal variants,
10:36
and their underlying mechanisms were unknown.
10:39
So using our California-based GWAS, we first looked in the Hispanic Latinos
10:44
and performed something called conditional association testing.
10:48
So what's that? You can see the top figure on the left-hand side.
10:51
This is just showing the p-values on the y-axis across the gene,
10:56
which you can see on the bottom there.
10:59
So these are simply the results from our GWAS, comparing the allele frequencies
11:04
between cases and controls.
11:06
Now, when we adjust for that most significant hit, if you look at signal 2 in
11:10
the middle there, you see the
11:12
second independent significant locus to the right-hand side of the gene.
11:16
And then when we adjust both for the signal 1 and signal 2 top snips,
11:20
we see this third independent risk locus at IKZF1.
11:25
And actually, interestingly, the two top hits in signal 2 and signal three are
11:30
only about 25 base pairs apart in the genome,
11:32
and they're both overlapping with a regulatory region or an enhancer region
11:37
that we hypothesized may impact expression of this gene.
11:43
Now, interestingly, when we looked in the non-Hispanic whites,
11:46
we only see two independent risk loci and we do not see that third signal.
11:51
So it appeared to be specific to Hispanic Latina populations, at least in our study.
11:57
So this novel variant has about a 20% risk allele frequency in Hispanic and
12:02
Latino populations, but is basically absent in individuals of predominantly European ancestry.
12:08
So it has a frequency of less than 1% in non-Hispanic whites.
12:13
So why does it have such a high risk allele frequency in Hispanic and Latinos?
12:17
Well, we found that the allele is actually positively correlated with indigenous
12:23
American ancestry in Hispanic and Latinos.
12:25
So you can see here with increasing global, so genome-wide ancestry proportions,
12:31
with increasing indigenous American ancestry, you see an increase in the risk
12:35
allele frequency of this SNP.
12:38
So why did it become so prevalent in Hispanic and Latino populations?
12:42
We next looked at ancient genomes to see when this allele may have arisen.
12:46
So we had access to this shotgun sequencing data in ancient DNA samples in the Americas.
12:52
And so the green here is showing presence of the IKZF1 allele,
12:56
and the darker the color, the older the individual that was sequenced.
13:00
So that red arrow is pointing to the oldest sequenced indigenous American from
13:05
almost 13,000 years ago.
13:07
And we showed that actually this individual was heterozygous for this leukemia risk allele.
13:12
And it's clearly present at quite a high frequency across the Americas.
13:16
So we next looked at why may this allele have become so prevalent? in the Americas.
13:25
So we explored signatures of positive selection at the IKZF1 locus and found
13:30
significant selection in two Hispanic and Latino populations,
13:34
the Peruvians in Peru and Mexican ancestry in Los Angeles.
13:39
We found no evidence of selection in a European population or in a East Asian population.
13:45
Even though this allele actually has a similar frequency in East Asians.
13:49
So we think our kind of running hypothesis is that the IKZF1 risk allele arose
13:55
sometime after the split of Europeans and Asians.
13:59
And then after the East Asians entered the Americas over the Bering Strait,
14:04
there was likely a reduction in the risk allele due to a population bottleneck.
14:08
And we think that there's since been selection leading to, actually in modern
14:13
day Indigenous American populations about a 50% allele frequency.
14:18
And then through admixture, we see about an 18% frequency in Hispanic and Latino populations.
14:24
I'd be happy to discuss the kind of potential reasons for the selection later on.
14:29
And so to summarize this part, we showed that Hispanic and Latinos are unique
14:34
in having three independent risk loci at this childhood ALL associated gene, IKZF1.
14:40
This allele has about a 20% frequency in Hispanic and Latinos,
14:44
is basically absent in Europeans, shows evidence of positive selection,
14:49
and we think contributes to the increased risk of ALL in Hispanic and Latino populations.
14:54
If you look in our manuscript, I didn't have time to go through this today,
14:58
but we performed some functional studies to try and understand the impact of
15:03
this variant on IKZF1 expression.
15:06
But we think, well, we know that the precise effect of this allele is still
15:11
to be determined. So we're currently working on.
15:14
Some cool functional studies in collaboration with Dr. Vijay Sankaran at Boston
15:18
Children's Hospital to precisely look at how does this allele impact the enhancer
15:24
region and its effects on IKZF1.
15:27
So what are we doing now?
15:30
I was fortunate enough to receive this R01 grant a few years ago to expand upon
15:35
our existing GWAS data sets in California.
15:38
So collaborating with groups in, well, Dr.
15:41
Shura, formerly at Texas Children's Hospital, now at Emory University,
15:44
Dr. Jun Yang at St. Jude's, and Dr.
15:47
Catherine Mattia at UC Berkeley, combining all of our data and generating new
15:51
data to perform the largest genetic study of ALL in Hispanic and Latinos.
15:57
So I wanted to highlight a recent paper that came out a couple of months ago
16:00
in HGG Advances, and this was work led by a graduate student, Jalen Lange.
16:06
So we're looking at local genetic ancestry. So I mentioned global or genome-wide ancestry earlier.
16:12
So local ancestry refers to the specific ancestral population that contributes
16:17
to DNA segments in an individual's genome at a particular location in the genome.
16:23
So we looked at all of the known ALL risk loci and investigated whether it was
16:29
present more often or not on the indigenous American haplotype.
16:33
And we found that actually for several of the known risk loci,
16:36
they were significantly more often on the indigenous American haplotype.
16:40
And actually, this was more often than not, increasing proportions of indigenous
16:44
American ancestry were associated with ALL risk at these loci.
16:49
We also performed something called admixture mapping, which leverages the admixt
16:54
genetics to look for loci that exhibit these frequency differences between ancestries,
17:00
and found four significant loci.
17:02
Two of these were already known, the ARID-5B and IKZF1, and there are two novel
17:07
hits on chromosome 2 and 15 that warrant further investigation.
17:12
I also wanted to highlight, I think, an important part of our paper,
17:15
which was comparing how self-reported ethnicity compares to infer genetic ancestry.
17:23
So we use this method called HAIR, a harmonizing ancestry and race ethnicity
17:28
that had been developed by other groups.
17:31
To see, and this HAIR algorithm, it's a special machine learning tool that categorizes
17:37
people into genetic ancestry groups for large scientific studies.
17:43
So for example, if in a study an individual is missing self-reported race ethnicity,
17:49
you could use this algorithm to predict their label based on the reference genome
17:55
populations and how their genetics correlates with those populations.
17:59
So we actually found that in our simulations, only about actually less than
18:04
two-thirds of individuals labels,
18:07
when we kind of forced the algorithm to think that everyone was missing their labels.
18:16
And when we tested using hair, less than two-thirds of individuals were recalled
18:20
as being Hispanic and Latino.
18:22
And of those that were not recalled back as Latino, about two-thirds of these
18:27
were assigned as being missing.
18:28
So we discussed the use of self-reported versus this hair assigned classifications.
18:34
Our view is that it's always better to use the self-reported labels rather than
18:39
trying to categorize people into different bins.
18:41
So if you have self-report race ethnicity available, that should really be your starting point.
18:47
Of course, you can consider genetic ancestry and how to account for that in your studies.
18:51
And we talk more about this in the discussion of our paper.
18:55
I was also fortunate to receive an administrative supplement from the NIH,
18:59
back when these were available, to expand our study to Hispanic Latino populations outside of the U.S.
19:06
So this is in collaboration with Dr. Silvia Jimenez-Morales and Dr.
19:11
Juan Manuel Mejia-Arangore at UNAM.
19:14
So we're genotyping additional cases and controls from a study from Mexico City.
19:19
Importantly, we actually have now several hundred cases from around Mexico,
19:23
from different regions. so we can really capture different ancestries across the country.
19:28
So I'm really excited about this study. We're also trying to use controls from
19:32
the Mexican Biobank study to increase sample size.
19:35
And we're also trying to pursue childhood ALK's control data from other studies
19:40
in Latin America, including Guatemala and potentially in Brazil.
19:45
And really, you know, as well as simply expanding sample size of our GWAS,
19:50
I think the really important part of this is that we can investigate the impact
19:54
of the known ALL risk loci and the performance of the polygenic risk score in
20:00
a Latino population outside of the US and potentially discover novel genetic
20:05
associations in these populations.
20:09
And so some ongoing projects. So Ms.
20:13
Elizabeth Quino, who's in the room here, is leading a GWAS of ALL as well as
20:17
childhood ALL patient outcomes in Mexican children.
20:21
And Benjamin Alonso, a student in our lab, is looking into the effective blood
20:26
cell traits in Hispanic Latino children on ALL risk.
20:30
Okay, so in my last few minutes, I want to briefly touch upon important research
20:35
from other groups that have looked at how genetic ancestry impacts somatic alterations
20:39
and outcomes in ALL patients.
20:42
So it was shown several years ago now that that Indigenous American ancestry
20:46
is associated with increased relapse risk in childhood ALL patients.
20:51
So this was first reported by Jun Yang's group. They also showed that a specific
20:56
SNP in the ARID5B gene was associated with increased relapse risk and that this
21:00
risk allele also correlated positively with Indigenous American ancestry.
21:05
So we think that as well as the ARID5B SNP contributing to increased ALL risk,
21:10
We also think it contributes to inferior outcomes in Hispanic and Latino patients.
21:16
And another important study that came out in JAMA Oncology three or four years
21:20
ago now looked at how genetic ancestry correlates with somatic alterations in childhood ALL tumors.
21:27
So it had previously been reported that there's a subtype of ALL called pH-like,
21:32
which has a gene expression pattern similar to the Philadelphia chromosome,
21:36
BCR-ABLE. That's why it's called pH-like.
21:39
So this subtype had been shown to be more prevalent among ALL patients of Hispanic and Latino ethnicity.
21:45
And in this study, they showed that genetic ancestry was correlated with this
21:54
pH-like, as well as the CRLF2 fusion.
21:57
And in a separate Guatemalan cohort, they showed a much higher frequency of
22:01
the CRLF2 rearrangement in individuals with high proportions of Native American
22:07
or Indigenous American ancestry.
22:09
And there's other paper that shows that the GATA-3 SNP may partially underlie this association.
22:16
Also important to highlight that both African and Indigenous American ancestry
22:22
remained associated with inferior outcomes even after adjusting for these molecular
22:27
subtypes and clinical features that suggests that there are perhaps additional
22:31
factors that correlate with ancestry,
22:33
for example social determinants of health,
22:35
structural racism, that also impact patient outcomes.
22:40
And just wanted to point out, there's another recent sequencing study in T-cell ALL.
22:44
In this paper, they also looked at genetic ancestry, somatic alterations, and patient outcomes.
22:49
Here, they did use this method of categorizing people into genomically defined ancestry groups.
22:55
One thing that's important to highlight is that 12% of the patients were actually
23:00
grouped as being complex.
23:04
Sorry, a team's meeting thing just popped up on here.
23:09
So these complex individuals didn't meet the ancestry threshold,
23:13
so they couldn't be binned into different categories.
23:15
So something I discussed in the commentary is how, you know,
23:19
when you group patients by ancestry, it's important to consider,
23:23
you know, how best to account for patients of mixed genetic ancestry.
23:27
And finally, I wanted to highlight that inferred genetic ancestry is only as
23:31
good as the reference populations that are used.
23:33
And so this was from the Mexico City Perspective Study, showing that when you
23:37
actually use more reference populations, you see more of a breakdown of different
23:42
ancestry groups, as you can see on the bottom here.
23:45
Okay, so our group and others have shown that common genetic variants contribute
23:48
to the increased risk of ALL in Hispanic and Latino children.
23:52
Several childhood ALL risk alleles have been positively associated with indigenous American ancestry.
23:58
Genetic ancestry can impact also the frequency of somatic alterations,
24:02
as well as their prognostic impact in both B-cell and T-cell ALL.
24:07
So important questions, how do we incorporate genetic ancestry into epidemiology
24:10
studies, clinical studies, as well as treatment decisions?
24:15
And it's important to expand our genetic analyses to populations outside of the US.
24:20
I don't mention it here, but also important to really understand what's causing
24:24
the increasing incidence of childhood ALL in Hispanic and Latino populations.
24:29
And finally, so how can we use this information for potential prevention of
24:34
childhood leukemia, especially in high-risk populations such as children of
24:38
Hispanic and Latino ethnicity.
24:40
So where we think this is going is one day, hopefully, we could be able to incorporate
24:46
the genetic data at birth, so identify everyone's genetic risk for childhood
24:52
leukemia as well as other cancers.
24:54
The possibility of screening for pre-leukemia opens up an avenue towards,
25:00
identifying those babies that may harbor these pre-leukemia clones at birth,
25:04
also looking at their genetic risk and also non-genetic or environmental factors
25:09
to really try and aim towards this precision prevention.
25:13
And so some of our ongoing research that I'm not talking about today,
25:16
but I'll highlight this record study where we're conducting backtracking in
25:21
cord bloods from kids who went on to develop leukemia.
25:24
And we're also starting to screen for pre-leukemia in unselected newborn cord blood samples.
25:31
And in my lab, specifically from a Hispanic-Latino population.
25:34
And I have collaborators who are looking at this in European ancestry populations.
25:39
So with that, I'd like to acknowledge all of my collaborators and look forward
25:44
to answering questions and more discussions. Thank you.
25:50
Any questions?
25:54
Thank you for that fascinating talk. I
25:57
wondered what the role of IKZF1 is in just its function and also its role in
26:09
ALL and whether it could be used potentially as a drug target for prevention or treatment of ALL.
26:17
I appreciate the pronunciation of the British Z there. So IKZF1 codes for Icarus
26:24
protein, and it's really crucial for lymphocyte development.
26:29
And so there are kids with rare pathogenic variants in the IKZF1 gene that have
26:36
immune dysfunction or immune deficiency.
26:39
Those same children or families also have an increased risk of ALL.
26:43
And so there's evidence that loss of IKZF1, and I should add,
26:50
somatic loss of IKZF1 is quite common feature of ALL.
26:54
So about 20% of the tumors actually have deletion of the gene.
26:57
And so we think that loss of IKZF1 leads to stalling of B cell development in
27:03
an earlier kind of pre-probie stage.
27:06
And those cells are propagating more in that early stage and are more likely
27:10
to develop the second hit mutations that cause leukemia.
27:14
So that some of our work is trying to understand first, what is the function of IKZF1?
27:19
And then what is the function of the risk alleles that may impact leukemia risk? Anybody else?
27:27
What is this? Oh, sorry. Did you have a question here? Oh, sorry.
27:32
You want to go first, Isabel?
27:36
What is the expected prevalence of this pre-leukemia? And, you know,
27:40
what sort of issues do you anticipate in telling someone your baby may be predisposed to leukemia?
27:46
So the prevalence of the pre-leukemia, is that the question? Yeah.
27:51
So for the ETV6-Rungs1 fusion, it's been found in up to 5% in published studies.
27:57
Actually, our collaborator is finding it even higher, maybe 10% frequency.
28:02
And that's just one subtype.
28:04
So we're also interested in branching out to look at other subtypes,
28:07
such as the CRLF2 fusion, which is more prevalent in Hispanic and Latino patients,
28:12
to see if we can also detect that at birth.
28:14
Yeah, but they're going off, I think there is, you know...
28:17
These students are telling a Latino family your child has leukemia. Exactly.
28:24
That's where I think the genetic risk will come into it.
28:28
If we can identify people in that top 5%, for example, of the polygenic risk
28:33
score and pre-leukemia, I mean, we'll obviously have to model this to see what is the risk.
28:38
Because, yeah, we do not want to tell parents that their children have pre-leukemia
28:43
without, first of all, without knowing what is the risk of leukemia.
28:46
Second of all, without an intervention.
28:48
And so that's also the missing part of this is what could we actually do?
28:52
Thank you, doctor. Great. This is really great. I've been following the,
28:56
you know, child pediatric leukemia for a long time with Latinos.
29:00
And I appreciate this because this is kind of where I was going to go with my question.
29:05
But as a community activist, not a scientist, to Luis's point,
29:10
how do we begin to intervene early enough to diminish that risk of late stage,
29:17
oftentimes with our Latino kids?
29:20
So where do we start and how do we start?
29:23
I remembered a presentation from Dr. Hillman back in 2009, when she was talking
29:28
about mutation in utero as a result of environmental exposure.
29:34
So how do we tie all of that together, start early education and intervention
29:38
and raising awareness for communities to diminish that risk for kids later on?
29:43
That's a really good question. I probably don't have enough time to answer it
29:46
properly. But I guess quickly, identifying environmental risk factors is difficult.
29:52
So we have some evidence of tobacco exposure being involved.
29:56
We are starting to conduct studies looking at environmental factors impacting
30:02
mutational patterns in the tumor to try and provide more evidence that they may be causal.
30:07
So I think it's really difficult in terms of the environmental factors and talking to parents.
30:14
Yeah, I think it's a really important issue and something to really think long
30:21
and hard about is how do we communicate this without scaring parents?
30:26
They might be worrying every time their kid gets sick, you know,
30:29
do I need to take him to the hospital?
30:31
But that's where I think we just need more.
30:34
The science isn't there yet, if I'm honest, in terms of that we don't have an intervention.
30:40
We don't know what the strong environmental risk factors are,
30:43
but we could tell parents their kid has an increased risk, but then what do
30:46
they do with that information?
30:48
Yeah, I'm happy to debate more over coffee or wine later.
30:53
It won't be, just a solution. Discuss.
30:58
Thank you. Thanks for this presentation. It was amazing. And,
31:01
you know, we keep reminding ourselves that, you know, we are trying to bring
31:05
knowledge and knowledge bring questions.
31:08
And then we have to have this way to deliver back to the community,
31:13
deliver back to the people who really want to have this knowledge.
31:15
But with that, we're going to continue the conversation with Dr.
31:18
Ruiz. Dr. Reese is an assistant professor of pediatrics in the University of
31:22
Pittsburgh School of Medicine and an attending physician in pediatrics and matology
31:26
oncology at the UPMC Children's Hospital of Pittsburgh.
31:30
She is a health service researcher.
31:32
Researchers investigate the social and structural drivers of health and how
31:37
they impact access, care, and outcomes in pediatric cancer.
31:40
She is establishing her research program focused on improving effective caregiver
31:45
clinician communications in pediatric oncology, just on topic.
31:49
The research has been funded by the Alex Eliminated Stand Foundation,
31:53
Conquer Cancer Foundation of ASCO, the Children Cancer Research Fund, and multiple others.
31:58
Please welcome Dr. Royce. Thank you.
32:04
Oops. Sorry. Oh, you're fine. Sorry.
32:09
Fix the microphone. Give me a second. All right. Thank you so much for the introduction.
32:14
I would also like to thank the conference committee for the invitation to speak today.
32:18
My talk today is titled Language and Outcomes in Pediatric Cancer Care.
32:23
And I agree that was a great segue in terms of the communication question.
32:27
I have no disclosures today. Brief outline of my talks.
32:30
I will start with my research journey and overall broad goals,
32:33
then move on to language and pediatric oncology background.
32:37
I will then present a qualitative research project and then move on to a quantitative
32:41
research project and end with a summary.
32:45
In terms of my research journey and goals. So my personal research journey,
32:49
so I'm a daughter of immigrants.
32:50
My parents grew up in central Mexico in a small little town where poverty was
32:54
rampant and they moved to Southern California in the late 80s.
32:58
That's where I was born and raised with my siblings and my extended family.
33:01
I grew up speaking Spanish as my first language and early in school I was in
33:06
English as second language classes.
33:07
My grandfather and also my extended family had renal disease,
33:11
and he ended up on dialysis.
33:13
And thankfully, he was able to get a renal transplant. And with that,
33:16
I was in high school, and I would often take him to the clinics.
33:19
And I saw that he was not offered an interpreter. And in fact,
33:22
the doctors asked me to be the interpreter, 16-year-old who had no medical knowledge or anything.
33:27
So you can imagine how that went. But this really was my first exposure to medicine.
33:31
And I thought, well, if this is medicine, I can probably do this better and have a bigger impact.
33:37
And so with that, I was pre-med. I was first in my family to go personally medicine
33:41
at Columbia University in New York City.
33:43
And it was during college where my baby cousin, this is a photo a couple of
33:47
years after she was diagnosed with infant leukemia.
33:49
I was in college, so every time I would go back to California, I would visit her.
33:53
And, you know, my aunt and my uncle, also primarily Spanish speaking,
33:57
I saw how there were certain doctors who were able to connect with them because
34:01
they were bilingual and how there were other doctors that were using very complicated
34:04
language and they would not understand And the treatment plan,
34:07
the long-term follow-up, and even now in survivorship, she is now 17,
34:11
so she is a survivor and a thriver.
34:13
But still, they have many questions about those long-term complications.
34:17
So my lived experience has influenced my broad research goals.
34:21
As mentioned, my goal is to evaluate the impact of social and structural drivers
34:24
of health on pediatric cancer access care and outcomes.
34:28
My focus right now is effective caregiver-clinician communication with interest
34:32
in families who speak languages other than English.
34:35
And the bigger goal is to develop multi-level interventions to make sure that
34:38
we're promoting equitable cancer care for all children.
34:42
That, I'm going to move on to the background on language and pediatric oncology.
34:45
So as mentioned in the previous lecture, despite advancements,
34:48
there are still ethnic disparities in cancer survival.
34:50
As mentioned, we know that ALL has the highest incidence in Hispanic Latino
34:54
children, and the disparities in ALL survival have been reported in the 1980s
34:58
and have continued and persisted in the 2000s.
35:01
Specifically, there's an 11 percentage point difference where Hispanic children
35:05
with ALL have worse survival compared to non-Hispanic white children.
35:08
So with these differences, you know, there's a really urgent need to investigate
35:12
other intersecting social drivers of health, such as language and English,
35:15
which is why I'm particularly interested in that communication between the family and medical team.
35:20
Some data in terms of linguistic diversity in the United States.
35:23
So we know that there are over 27 million people that identify as limited English
35:27
proficiency in the U.S. census.
35:29
Of these, 63% speak Spanish as the primary language.
35:34
And we also know more recent data that more than 26 percent of children in the
35:38
United States are children in immigrant households.
35:42
What are our language access laws and policies? So going back to the Civil Rights
35:46
Act, this prohibits the recipient of federal funding from discriminating against
35:50
individuals because of race, color, or national origin, and this incorporates
35:53
language in that part as well.
35:55
And more recently, Section 1557 of the Affordable Care Act reinforces these
35:59
language services as a fundamental right.
36:01
Specifically, the language-assisted service requirements need to be free of charge, accurate.
36:07
Protect privacy and independent decision-making of the individual with LAP, and also, importantly,
36:13
cannot rely on staff other than qualified interpreters and also prohibits the
36:18
use of minors and or accompanying adults as interpreters or translators except in emergencies.
36:24
So how are we doing in pediatric oncology with language access?
36:28
Here I bring up two studies on clinical trials and language in pediatric oncology.
36:34
I did a study looking at pediatric phase one and two studies that are listed
36:38
on clinicaltrials.gov and only 4% reported, explicitly reported language requirements.
36:45
And I was involved in another study with the Children's Oncology Group Language Equity Working Group,
36:50
where we surveyed all the COG sites in the United States, and 47% of the respondents
36:56
reported that somewhat or very difficult to recruit patients with LAP or speak
37:01
other languages other than English.
37:02
There is an older study back in 2003 where they actually used audio recordings,
37:08
for newly diagnosed BALL, and they had Spanish-speaking families and English-speaking families.
37:13
And from those audio recordings, we learned that oncologists provided leukemia
37:17
treatment details only 65% of the time for the Spanish-speaking families compared
37:21
to 71% of the time for the English-speaking families.
37:24
And we also, from these audio recordings, learned that 50% less questions were
37:29
asked by parents who, in this case, were Spanish-speaking compared to the English-speaking parents.
37:35
There are other studies looking at challenges in cancer care.
37:38
So a survey from 2006, they found that Spanish-speaking caregivers reported
37:42
comprehension difficulties related to the interpretive process.
37:45
And then 10 years later, a different survey, more than 25 percent of the Spanish-speaking
37:49
caregivers reported difficulty communicating with physicians,
37:51
and over 30 percent reported difficulty communicating with support staff, such as nurses.
37:56
So with all this data, I was very interested in trying to capture more in-detailed
38:01
information from the actual Spanish speaking caregivers that maybe we could
38:07
not capture from these surveys.
38:09
And that takes me to the first study that I will talk about as my qualitative research study.
38:13
So the objective was to understand Spanish speaking caregivers' communication
38:17
experiences with the medical team at the time of that new diagnosis and to identify
38:21
strategies to improve communication.
38:23
My growing research team here included Elizabeth Contertas, who is a graduate
38:27
student in the education program at the University of Pittsburgh,
38:29
and Alejandra Corbello-Paz, who is now a third-year medical student at the University
38:33
of Pittsburgh School of Medicine.
38:35
Both of them are Latina and bilingual, which is a huge asset to my research team.
38:39
So the significance of this project really our hope is to identify areas of
38:42
strengths and opportunities related to effective caregiver clinician communication
38:45
directly from the Spanish-speaking caregivers.
38:48
The inclusion criteria included Spanish-speaking caregivers of children with
38:51
cancer who were diagnosed in the last 10 years, either where I work at the Children's
38:55
Hospital of Pittsburgh or where my collaborator works at the Children's Hospital
38:59
of Orange County in California.
39:01
Caregivers needed to be adults and needed to be present during the initial oncologic
39:05
diagnosis conversation.
39:07
So the interviews were conducted by my bilingual research team.
39:10
They were all recorded and transcribed either in phone or in person,
39:13
and the participants received a gift card at the end for participating.
39:17
Some of the topics of the interview guide, the perceptions of that initial diagnosis
39:21
conversation, methods for methods participants sought information regarding
39:25
the diagnosis outside of just the
39:27
doctors, and suggestions for improving communication for future families.
39:32
Terms of data analysis, a preliminary codebook was created inductively.
39:36
Transcripts were analyzed by two coders separately and then met to compare the codes.
39:40
We kept refining the codebook systematically and iteratively.
39:43
Once the thematic saturation was reached, we developed a final codebook,
39:47
which was then reapplied to all the transcripts.
39:49
We identified themes and patterns, and we used NVivo to store and organize all the data.
39:53
So we have, in terms of results, we did 11 interviews.
39:57
100% of them were with the mothers. The average age of these mothers was 41,
40:02
although there was a range of 22 to 56.
40:05
Not surprising, the most common cancer diagnosis of the child was leukemia at
40:08
73%, although we did have also sarcomas and brain tumors.
40:12
The most common country of origin of the caregiver was Mexico at 82%.
40:16
In terms of the average years in the United States, that was 17,
40:19
but also a very broad range between 2 to 36.
40:22
And the highest level education was a high school graduate or less at 91%.
40:27
So I'm going to highlight a couple of themes. One of them was variable interpreter
40:30
access, as well as the emotional response to diagnosis.
40:34
Additionally, divergent comfort levels with asking questions.
40:37
And finally, how families are establishing therapeutic relationships with the medical team.
40:42
So I'm going to highlight some key quotes that really talk to these themes.
40:47
So I have a couple for variable interpreter access.
40:50
So even in the 2000s, 2020s, families being used as ad hoc interpreters.
40:56
So here the mom said, my sister, she speaks Spanish and English.
41:01
They told her we found cancer.
41:05
Here's another example where the mom didn't find any bilingual staff within
41:09
the hospital, but once they moved the child to the ambulance,
41:12
she saw that the ambulance, one of the ambulance drivers was bilingual.
41:16
So she said in the ambulance, the man in charge, he spoke Spanish very well.
41:20
And I asked him if he could give me an idea of how serious this was.
41:25
So you can just imagine how, you know, they've been told you have to be transferred
41:28
to a pediatric hospital because there's something wrong and she doesn't understand.
41:32
Answer. She's asking the ambulance driver.
41:35
There was also several instances where the actual patient who,
41:39
this teenager, so a 17-year-old who's bilingual, is being asked to interpret
41:44
his own abnormal blood results.
41:46
So the mama is explaining how the doctor was saying, I found something in your
41:50
blood and I don't like it, she tells him.
41:53
I'm very surprised at the level you have. It's a lot.
41:56
The interviewer asked the mom, is the doctor speaking to you or more to your
42:00
son, and in English or Spanish.
42:03
So then the mom says the doctor was speaking to him in English.
42:06
So again, if you remember from the previous slides, this is not supposed to be allowed.
42:11
The patient, a minor, should not be interpreting unless it's an emergency.
42:15
And the fact that he's talking makes me assume this is not an emergency.
42:19
There was also several instances where caregivers mentioned incomplete interpretation.
42:23
So here a mom says we have conversations that they also, So we have conversations
42:28
that they also translate both in person or in videos.
42:31
Then my son tells me, Ma, they didn't tell you everything.
42:35
While they translate it, I don't know if they're telling you everything.
42:38
So I think as a lot of clinicians, we know that sometimes the interpreter is
42:42
not saying everything either from the clinician or from the family.
42:45
The families are also picking it up.
42:48
Emotional response to diagnosis was seen in all the interviews.
42:50
A lot of them spoke about their own emotions as caregivers.
42:54
So here, one caregiver says, when they gave us the news, we felt really bad
42:58
because in my country, when we hear about this disease, for us,
43:01
it means that it is no solution. A couple also mentioned the child's emotions.
43:05
So here a caregiver says, she heard it, that she has cancer.
43:08
She cried. And from there I saw the agony began.
43:12
There was a lot of instances where caregivers brought up asking questions.
43:17
Here's a quote where one caregiver said, sometimes I felt embarrassed to ask
43:21
questions because I didn't know English.
43:24
Here, one says, I know they explained to me, but I had more questions,
43:28
yet because of time, because of the emergency, I couldn't ask any more questions.
43:34
Here's a very powerful quote where the caregiver describes kind of the clinician
43:39
giving them permission to ask questions.
43:41
So she states, they were very good because they told me, you can ask us as many
43:45
times as you want, any questions you have, ask us.
43:48
We will answer them for you five times, ten times, as many as you have,
43:52
and we will answer them for you, and well, the truth is, they were very kind.
43:57
And along those lines, in terms of establishing those therapeutic relationships
44:01
with the medical team, another caregiver said, We saw how they cared,
44:05
how they attended to us, and above all, the empathy. That is very important.
44:10
Caregivers also provided several recommendations for us to improve that communication
44:15
at the time my diagnosis.
44:16
They want more bilingual clinicians. They want in-person interpreters.
44:20
They want detailed explanations, and they want the medical team to take perspective taking.
44:26
So some illustrative quotes. So if we speak Spanish, find doctors who speak
44:30
Spanish so that one can understand.
44:33
Try to find an interpreter who speaks
44:34
Spanish or a doctor who can give the diagnosis in the same language.
44:38
In terms of quotes for detailed explanations, that they tell me but explain
44:42
well, look, your daughter has this and this, but we will resolve this with the
44:46
medication or we will do this.
44:48
Another caregiver said, I think that sometimes there's a need for a group where
44:52
parents, we are offered a little
44:54
more information in our language with simpler terms to understand it.
44:59
And then perspective taking, a caregiver said, because sometimes I do feel that
45:03
it is also something good for doctors to put themselves in the parents' shoes,
45:06
because when someone is going through that, it is very hard.
45:10
So some conclusions from this qualitative study. So again, even in the 2020s,
45:15
caregivers experienced variable interpreter access and accuracy.
45:19
There's a very strong emotional response with the new pediatric cancer diagnosis,
45:22
and the medical team needs to know how to acknowledge it and address this.
45:26
And the oncologists need to encourage and normalize question asking for all families.
45:31
And obviously, there's still a very huge need for bilingual skills in medicine.
45:35
With that, I'm going to move on to a different study, a more quantitative research project.
45:39
So the specific aims for this study was to examine the association between preferred
45:43
language and high acuity in children with acute myeloid leukemia at time of
45:47
diagnosis at the first presentation to the hospital.
45:50
And the second aim was to examine the association between preferred language
45:53
and intensive care unit or ICU level care during the first cycle of therapy
45:58
in children with AML, knowing that all of these children are hospitalized for that whole first cycle.
46:04
So that kind of takes away the access to health care into the equation.
46:09
And I hypothesize that speaking languages other than English for this study,
46:14
specifically Spanish, would be associated with high acuity representation as
46:17
well as ICU level care during the first cycle.
46:19
This study was funded by the Concord Cancer ASCO Foundation through Young Investigator Award.
46:23
So the data source is the real AML cohort.
46:27
So pediatric AML is a very rare disease. So, for example, in Pittsburgh,
46:31
we have over 200 cancer diagnoses.
46:33
But only three or four are pediatric AML. So we can't really do single-center
46:38
studies, so we need to collaborate with other institutions. So when I was in
46:42
Philadelphia for my fellowship, Dr.
46:44
Richard Afflitz was one of my mentors, and he, along with Dr.
46:48
Kelly Goetz, created this real AML cohort where multiple institutions are contributing
46:53
the patient's data for us to then be able to ask really rigorous epidemiological research questions.
47:00
So from my study, it was eight institutions who contributed data and my data
47:04
set was from January 2011 to October 2023.
47:09
Bit more details about how we get this data from the different institutions.
47:14
So again, multiple institutions, they identify the patients who have AML,
47:18
and then they send that information to the Children's Hospital of Philadelphia,
47:22
University of Pennsylvania.
47:23
There's a very rigorous training to abstract the data and make sure that there's
47:29
quality testing for this abstraction.
47:31
And then there's a lot of data quality checks, as well as data issue resolutions.
47:36
The team then freezes the data at different intervals. And then this allows
47:40
us to have an extended data linkage to other national databases and then be
47:46
able to then make multiple analytic data sets depending on the questions we have.
47:50
So for my study, my primary exposure was a combination of race and ethnicity
47:54
along with the preferred language.
47:56
And my two outcomes that I looked at was high acuity, which we defined as having
48:01
at least two or more organ systems requiring that ICU level care during the
48:06
first 72 hours of that initial admission.
48:09
And then my other outcome was any ICU-level care during that first cycle of treatment.
48:15
Covariates that I looked at included age, sex, insurance, year of diagnosis,
48:20
risk classification, as well as trial enrollment status.
48:23
So in terms of statistical analysis, the covariates were compared by exposure.
48:28
I performed unadjusted and adjusted regression models, either logistic or log
48:33
binomial, depending on our outcome.
48:35
And then all of this was done on Stata.
48:38
For the results, so the entire cohort, I don't.
48:42
Had 479 AML patients. Of these, 210 were non-Hispanic white English-speaking,
48:50
which was the preference group for this study.
48:52
And then the population that I was interested in are Hispanic Spanish-speaking
48:55
patients. We had 50 of those.
48:58
We also had 79 who were Hispanic English-speaking patients.
49:03
And then we looked at non-Hispanic Black English speakers, which was 86,
49:07
and non-Hispanic other English speakers, which was 55.
49:10
So then we looked at by the age when they were diagnosed by all these categories.
49:15
So there wasn't any major differences
49:18
between the age at that time of diagnosis for these categories.
49:22
We also looked at sex. So whether we saw that for females, there was an outlier
49:28
where the Hispanic Spanish speaking group, there were less females compared to the other groups.
49:33
We also looked at insurance, whether they had any private insurance or only public insurance.
49:39
This was also, we found a statistical significance between the groups.
49:43
And to take that into our analysis, I pulled out only the patients who had public
49:50
insurance, since we know oftentimes that can be a proxy for access to health care.
49:56
And then I looked at trial enrollment, which was looking at to see if they were
50:00
enrolled on an open clinical trial during their treatment or not.
50:03
And then I also looked at years of diagnosis, which was not different between the groups.
50:08
And finally, we also looked at risk classification, whether they were low-risk,
50:12
AML, intermediate, or high.
50:13
And this also, we did not find any statistical significance between the groups.
50:18
But when we look at our actual outcomes of interest, so this is looking at the
50:23
point estimates of those outcomes for only the kids that have public insurance.
50:28
So for the high-acuity presentation, that first 72 hours, we see that 16% of
50:35
the Hispanic Spanish-speaking patients are presenting with high acuity, which is...
50:42
Double digits, and you can see from the other groups, they're all in the single digits.
50:47
And then our second outcome was any ICU-level care during that first cycle of treatment.
50:52
And again, you can see in our Hispanic-Spanish-speaking group,
50:55
it's at 54%, where the rest are around the 30s, with the exception of the non-Hispanic
51:00
other English-speaking.
51:02
So these are just the point estimates. And then I did the regression analysis.
51:08
So going back to our first outcome, high acuity at presentation for patients with public insurance.
51:14
You can see that in an unadjusted as well as an adjusted models,
51:18
Hispanic, Spanish-speaking patients had higher than four times higher odds of...
51:26
Having high acuity at presentation compared to our non-Hispanic white English-speaking patients.
51:31
And the rest of the groups did not have, and that was statistically significant,
51:36
while the rest of the groups, we did not have statistical significance.
51:39
And then when we looked at ICU-level care for patients with public insurance,
51:44
we also, for the Hispanic Spanish-speaking, we saw that in our adjusted model
51:49
for adjusted for age and sex, the risk ratio was almost two times higher for
51:54
Hispanic Spanish-speaking patients compared to their non-Hispanic white English speaking.
51:58
And again, this was statistically significant while the rest of the other groups were not.
52:03
There are, of course, limitations to the study. You know, language is captured
52:07
by the electronic medical record, and that's per individual institutional standards.
52:13
There is no information in this study regarding access to medical interpreters
52:17
or use of medical interpreters.
52:19
And as mentioned before, language is only one of many social drivers,
52:22
which in combination with biology need to be further investigated.
52:25
To understand and intervene.
52:27
So conclusions of this quantitative study, compared to non-Hispanic white English-speaking
52:31
patients, Hispanic Spanish-speaking AML patients are both significantly more
52:35
likely to have high acuity at presentation and higher ICU-level care needs.
52:40
Interestingly, Hispanic English-speaking patients did not show this difference,
52:44
supporting our hypothesis that speaking language other than English,
52:47
in this case for the study Spanish, is a risk factor for early morbidity.
52:52
We clearly need more research on interpreter use and family-centered communication
52:56
for oncology families that speak languages other than English.
52:59
To summarize the two studies, so language access is mandated by law,
53:04
and I think we can all try to be advocates in our clinics and our hospitals
53:08
and institutions and our companies to ensure that we are providing interpretation
53:13
and translation for all our families.
53:15
Outcome disparities based on
53:16
language and other social drivers of health exist in pediatric oncology.
53:20
And finally, we should strive to partner with medical interpreters for families
53:23
who speak language of Southern English for every clinical encounter for cancer care.
53:26
And with that, I'd like to acknowledge all my mentors, both in Pittsburgh and
53:30
in Philadelphia, as well as my growing research team and all the funding support
53:33
that I've been fortunate to have.
53:36
And finally, huge thanks to the patients, caregivers, and families who participate
53:39
in our studies. I'm happy to take questions and also move on to the panel part of this.
53:49
Hi, thank you. This question is actually for both doctors.
53:53
So I'm a survivor. I'm a patient advocate. I'm a survivor for AML.
53:58
And the fact that you both are targeting AML, is there any statistics or studies
54:04
that you have done for Latino patients that their parents had AML, but it's not heritary?
54:12
Is there something out there for Latinos that they are aware that even though
54:18
the parents had AML, they are not going to get it or a third or four generations can get it?
54:24
Is there any literature or any statistics for them?
54:28
And the second question would be, if there is, what can we as patient advocates,
54:34
how can we communicate this to our communities?
54:41
I'm not a medical doctor, so I differ. I can take a stab at it. Sure.
54:48
So the majority of pediatric AML is extremely rare and is not inherited.
54:53
There are certain genetic predispositions where the gene is passed down from
54:59
family to family member to family member.
55:01
And that's a very different population compared to the majority of pediatric
55:05
AML. In terms of specific studies within the Latino population,
55:09
I don't know any at the top of my head.
55:12
But for pediatrics compared to adult cancers, most of it is extremely rare and
55:17
is not passed down. It's not inherited.
55:20
Thank you. And the reason why I was saying is because the more I advocate for
55:24
leukemia, for AML, and within our community, there's not enough.
55:29
And this is probably raising my voice saying that if you guys are doing for
55:34
children, to raise it up.
55:36
Because it's insane how the more I speak on it and the more I approach, there's barely few.
55:43
And even to explain to a child that they had AML.
55:47
The literature is very limited in our community.
55:54
So thank you. Thank you for sharing that. I completely agree that we need to do more studies.
55:59
Hi, Ian. Thank you, guys, Ian. And my question was if there's any data that
56:05
can compare, you know, patients that might have missed appointments,
56:09
treatments, or even the choice of treatment, depending on, like, the stage.
56:14
Like people with language barrier or not
56:17
like limited access to interpreters from those
56:20
that you know do speak english yeah yeah no
56:23
i do think a lot so once you get into the
56:26
system of a pediatric oncology hospital it's extremely
56:30
rare that you will not like you will come to the appointments because your life
56:33
depends on it and everyone there's a whole large multidisciplinary team to ensure
56:38
that if you don't have a car we'll find transportation for you so all of that
56:43
is set up in the pediatric cancer center once you get into the system.
56:46
I think you're also kind of alluding to what about before, right?
56:50
So are the patients going to their pediatrician on a regular basis?
56:56
Are they having, when they get sick, are they going to see their doctor when
57:00
they're getting sick now four or five times?
57:02
Clearly something is wrong and maybe we should start thinking about cancer as a possible diagnosis.
57:07
So that's pre the actual cancer diagnosis and I think also plays a huge role
57:11
in terms of other social drivers of health and making sure like mentioned earlier
57:15
not getting super super sick that now you end up in the icu when we could have
57:20
caught it earlier if they had seen their doctors and someone had done the blood
57:23
tests earlier absolutely.
57:26
Forget the second part was something about interpreters well yeah so yeah just
57:31
comparing yeah so like the people who do not have yeah access to the interpreter
57:37
no that's a great question,
57:40
is actually a feature direction of some of my studies to capture.
57:43
We actually don't know for families who don't speak English,
57:46
what types of interpreters are being used. Are they in person?
57:49
Are they over the phone? Are they the video?
57:51
And so one of my ongoing studies is to actually capture that data.
57:55
Who are we using for interpretation?
57:57
And what training do they have? What experience do they have within pediatric oncology?
58:01
Because I think all of that will play a role into how they become part of the
58:05
medical team and interpret for the families and the doctors. Absolutely.
58:10
Thank you. I want to just say thank you to everyone also, because I clearly
58:13
have a child who wanted to had questions
58:15
in the middle of both of your presentations. So sorry about that.
58:18
I wanted to come back because I truly valued what you found in the 2020s,
58:23
since we know so much in that qualitative study has been repeated and repeated in there.
58:28
And I was wondering if you would be able to share a little bit about the facility context, right?
58:34
Because so often we focus on the clinician, but who the clinician,
58:38
who hires the clinician, right?
58:39
And who hires the team and whether it's, yo lo sé papi, deme.
58:43
I think that where I'm getting at here, right, is that non-academic safety net
58:47
systems, they actually have a full variety of multilingual teams that are available.
58:52
What they don't have is necessarily access to clinical expertise to be able
58:56
to do that specialty care as well.
58:58
Right. And so really this conversation around linguistic disparities to your
59:02
interest and your passion around structure really incorporates the facility.
59:07
Like, does the facility actually have a barrier, represent barriers linguistically,
59:12
or is it a facility problem? That's really my question, part one.
59:16
And then, two, and I know you get this often, so I'm just going to throw it
59:19
out there, but just putting it out there.
59:20
I think in line with that, the facilities that are highly resourced,
59:25
right, because of the patient population that they naturally have,
59:29
which influences the kind of clinical representation they have for trials, right,
59:32
they potentially value less interpretation in terms of using phone-based interpreters,
59:38
right, or using the dial-in number or things like that because that's cost-effective for them, right?
59:42
And that means that they rely specifically on exact translation.
59:47
Which we know is a challenge. We know the value of exact translation.
59:50
We know the problem. Our metaphors are different. Our analogies are different.
59:53
How we communicate, even if you're talking at a fifth grade level in English, it is challenging.
59:58
Martian, right, in Spanish, right? So I just, I would love your reflection on
1:00:02
the facility context where you're leading your past work, but your current work,
1:00:06
and how you're incorporating these facility level challenges,
1:00:09
right, that really then feed into the relationship and then influence really
1:00:13
the workforce that we need to be investing in.
1:00:16
Pick one of those things, whatever that is.
1:00:19
No, absolutely. I do think, like you mentioned, facilities, the country is very
1:00:25
heterogeneous, The resources are very heterogeneous.
1:00:28
I've now been at three hospitals, so I've actually seen it, you know,
1:00:31
from very urban to now have a population where it's mostly rural.
1:00:35
My current institution actually does not have a language services department,
1:00:39
right? We don't have in-person interpreters.
1:00:41
We have to make an appointment with them from an outsourced company for them
1:00:45
to then come in versus my other institutions. They have in-person interpreters.
1:00:50
They're already on site. It's easy to call them. So yes, I think that plays a huge role.
1:00:55
And I think, you know, I'm starting the study in Pittsburgh,
1:00:59
but I'm hoping to then be able to move it to multi-sites to really get into
1:01:02
that in terms of the types of interpreters.
1:01:05
And maybe we don't need that in-person interpreter for everything,
1:01:08
but we don't know that yet, right?
1:01:10
So I'm hoping to be able to capture that data, to be able to make recommendations based on evidence.
1:01:15
And then in terms of resources, absolutely.
1:01:19
I know that some really high-resourced children's hospitals,
1:01:24
they have the money to have the in-person interpreters and even dedicated pediatric
1:01:29
oncology interpreters. So that is their life and world.
1:01:32
They know the lingo. They've done it so many times.
1:01:34
I would love for all children's hospitals to have that, right?
1:01:38
But we can't. They don't have the resources.
1:01:40
But what can we learn from them,
1:01:41
that they're doing well, that we can potentially adapt for other sites?
1:01:46
Is there somebody up here? Yes.
1:01:50
Thank you both for the amazing presentation. So I guess first I have a question for Dr. Ruiz.
1:01:55
So when we're thinking about language and I guess like no pun intended,
1:01:59
some of the language that you had in the last slide mentioned language as a
1:02:02
driver of social determinants to help but also as a risk factor.
1:02:06
I'm just curious, what do you think that we're actually capturing when we say
1:02:10
we have an Hispanic and Latino that only speaks Spanish?
1:02:13
Like what else comes with that label that we're assigning?
1:02:17
We've done analysis looking at nativity, for example, looking at ethnic and
1:02:20
click resonance, thinking of if we were to change, have the interpreters available,
1:02:26
what else is it that that variable is capturing that we would need to intervene
1:02:30
to actually improve the outcomes of that specific person.
1:02:34
If I understand your question correctly, essentially, what other risk factors
1:02:39
or drivers do we need to take into account, especially for families who don't
1:02:42
speak, who are Spanish-speaking, Latino? Is that right?
1:02:46
Yeah. What else is associated or correlated with that specific exposure? Okay.
1:02:53
Because just by changing the language alone, it might not be enough to actually impact the outcome.
1:02:58
Right, right. And, you know, with Dr. Dismith's talk, there's a lot of genetics
1:03:02
involved as well. I'm not saying that only changing the interpreter access will
1:03:07
change outcomes. Obviously, we need to have multilevel interventions coming
1:03:11
from different directions.
1:03:13
Some of my studies are capturing, you know, oftentimes we think about other
1:03:16
social drivers, so poverty.
1:03:18
You know, I have families who don't have cars, who are like single mothers.
1:03:25
So that information we need to capture so the medical team knows,
1:03:29
so our social worker knows.
1:03:30
You know there's a public program that we
1:03:33
can book you transportation so that's
1:03:36
not an issue for you to come to clinic right so i think that's a lot we should
1:03:41
strive to capture a lot of socio-demographic data to be able to then say well
1:03:45
it's not just a spanish-speaking latino family but is it that they need financial
1:03:49
support in other ways as an example i believe you have an answer,
1:03:55
Could I ask that? Yes, please.
1:03:57
I'm sorry to interrupt you last. No, can I quickly follow on your question? Yes, please.
1:04:04
This is also getting back to your earlier point. I had assumed,
1:04:08
probably because I'm not a medical doctor, that with acute leukemia,
1:04:11
when the symptoms start to show,
1:04:14
pretty quickly the parents would realize something is really wrong and they
1:04:18
would take their child to the doctor.
1:04:20
Is there any evidence that, for example, the Spanish-speaking Hispanic Latinos,
1:04:25
there could be factors that cause them to delay going to the doctor?
1:04:29
For example, immigration issues, right? Fears of, and could that lead to the
1:04:33
higher acuity issue? Right.
1:04:38
So I would say in general, for any family, when your child gets sick,
1:04:43
you're going to be told it's a virus.
1:04:46
They're fine. Give them Tylenol. And 99% of the time, that's right.
1:04:51
You know, they get a lot of illnesses. It's part of being a child.
1:04:54
It then becomes the point where, so sometimes these families have multiple doctor
1:04:58
visits, multiple ED visits, urgent care visits, before they get to the cancer
1:05:03
diagnosis, because it is so rare in children.
1:05:06
So that's normal across any type of family, right?
1:05:09
But as Dr. Smith mentioned, maybe for Latino families, you know,
1:05:14
they don't have an established primary care doctor. So they are not going.
1:05:18
So they're just at home and they're doing, you know, Vaporul, Tylenol, Ibuprofen.
1:05:23
But they don't have that established relationship with a doctor to be able to
1:05:27
go and, you know, be examined and notice things sooner compared to a family
1:05:31
who does have that doctor. and they're like, oh, you're actually really pale
1:05:34
or actually your heart rate's really high. Well, actually your spleen is really big.
1:05:38
So those are things that they would pick up on if they're being examined by
1:05:41
the doctor that would potentially lead to a cancer diagnosis more rapidly than
1:05:46
families who don't have that doctor. So absolutely.
1:05:48
So maybe that's where education could come in.
1:05:51
Yeah, thank you. So for Dr. Smith, summary to you, as you mentioned,
1:05:55
you were building polygenic risk scores.
1:05:57
I guess the first question, where are the weights coming from to build those
1:06:01
polygenic risk scores that then you're going to deploy in Hispanic and Latino populations.
1:06:05
And then from what I recall, top of my head, there's very few genes that have
1:06:09
been shown to have a genome association with leukemia.
1:06:13
I think it's like less than 30, something like that. So how much does the predicted
1:06:17
ability of the score changes once you add variants that seem to have a higher
1:06:20
allemic frequency to specific populations as the one you're working with?
1:06:27
Yeah. So in terms of building the scores, you know, we can typically take,
1:06:31
if the study is large enough, we can split it into the testing,
1:06:36
validation, replication phases.
1:06:39
What's complicated about the Hispanic Latino population is the heterogeneity.
1:06:43
So that's why it's important for us to expand our studies to populations outside
1:06:48
of the U.S., I think as well as testing in populations across the U.S.
1:06:53
So a polygenic risk score developed in California is not going to work the same
1:06:57
for kids in New York or Florida, right?
1:07:01
So yeah, that's definitely a work in progress in terms of how do we determine
1:07:06
the best polygenic risk score.
1:07:08
It might turn out that a multi-ancestry PRS that actually could work for all
1:07:14
kids might be the way forward, incorporating some of the population-specific
1:07:19
alleles like IKZF1 as well as for other populations.
1:07:23
So does that answer your question?
1:07:26
Yeah. We can talk about it. Sure.
1:07:29
Oh, thank you.
1:07:33
I'm going to go to left field on this, because I've heard this over and over
1:07:37
for the last two years. AI.
1:07:40
UCSF did a study survey with Latino patients, half of them English-speaking
1:07:46
and the others, and for the most part, they said, we heard about this AI.
1:07:52
Majority of them said, I don't want a robot to decide my treatment. That's very basic.
1:07:58
How do you, do you use AI?
1:08:01
How do you use it? Will it move, you know, findings quicker that make a difference,
1:08:08
a positive difference and make a difference in Latino lives?
1:08:12
Or do you need to create some primers and teach patients about it,
1:08:18
how to look for it in their own treatment plans, how to add information that's critical.
1:08:25
I mean, there are two sides to this coin, but I'm wondering how all you are
1:08:28
addressing this issue or if
1:08:30
it's going to make an important positive difference in what you're doing.
1:08:34
I think just very quickly, we are considering using certain tools for genomic
1:08:40
analyses to look for patterns of somatic mutations.
1:08:43
But in terms of treatment plans at NFC.
1:08:48
Yeah, so thankfully for pediatric cancers, the treatment plans are extremely
1:08:53
standardized across the country and across the world, honestly.
1:08:56
So we either use standard of care protocol, which everyone is using,
1:08:59
or there's an open clinical trial that has a research question to it,
1:09:04
and they might be eligible for that.
1:09:05
So I would say the majority of pediatric oncologists are not using AI for upfront treatment decisions.
1:09:14
Now, when you're getting to relapsed, first relapse, second relapse,
1:09:18
relapse refractory, now we know that our standard treatments are not working.
1:09:23
Maybe we can use AI to see what's out in the literature, right?
1:09:26
But now we're talking about case reports, like two patients here did this,
1:09:30
five patients here did that, and these are the results.
1:09:32
So I can definitely see AI being helpful in those situations where we can gather
1:09:36
data or gather useful, potentially, publications for us to then read it in depth
1:09:42
to see if it's relevant for that patient.
1:09:44
I always get the AI question when it comes to interpretation and translation
1:09:47
because I do think, you know, in the future, it might be a positive thing to add.
1:09:53
Obviously, it needs a lot of training to be able to do the translation interpretation
1:09:59
for maybe some routine things like maybe the nurse is asking, do you have pain?
1:10:03
As opposed to having a whole new diagnosis cancer conversation with AI that
1:10:08
I think we will always need a human for that.
1:10:13
Wondered whether there are any clues. I mean, a lot of these studies are still
1:10:17
quite small, I mean, in terms of the number of cancer cases.
1:10:21
I wonder if there are any clues that could be drawn out from how they've approached
1:10:25
this in adult diseases such as cancer,
1:10:28
such as heart disease, in either developing a PRS or developing a language-based
1:10:35
approach to interpreting or giving recommendations.
1:10:41
So I just know that breast cancer, for example,
1:10:46
heart disease have much greater PRS, which has been associated with the numerous
1:10:53
loci, which seem to have cropped up recently from GWAS.
1:10:58
Yeah, one difference is that the adult cancers, typically the effect size of
1:11:03
those variants are quite small compared to childhood cancers tend to have a stronger effect.
1:11:09
We think, you know, because in adult cancers, they've had a lifetime or a longer life of exposures.
1:11:15
So the PRS for childhood ALL may be just as powerful with 25 SNPs as a breast
1:11:20
cancer PRS with 300 SNPs.
1:11:23
But no, we definitely do. We are inspired by those larger studies in terms of
1:11:28
some of the things that we do, you know, oh, you know, we see a paper in prostate
1:11:33
cancer, we can apply that to childhood ALL.
1:11:35
So, yeah, it's definitely a good point.
1:11:39
Well, I'd like to thank you all for being here and asking really great questions
1:11:42
and for the classical thoughts.