European Association for the Study of Diabetes (EASD)

September 14-19, 2014: Vienna, Austria – Full Report – Outcomes Trials

Executive Highlights

The conference featured noteworthy commentary and discussion on cardiovascular outcomes trials (CVOTs), as speakers highlighted their limitations and the need to set realistic expectations about what they will show. AstraZeneca’s Monday symposium featured perhaps the best CVOT-related talk of the conference. Dr. Stephen Gough (University of Oxford, Oxford, UK) pushed attendees to set realistic expectations for ongoing CVOTs, explaining that the studies currently enroll later-stage patients for whom the “window” for cardioprotection is nearly closed. To get more out of future CVOTs, he suggested longer studies that recruit patients soon after diagnosis and with minimal established cardiovascular risk. More specifically, Dr. Robert Gilbert (University of Toronto, Toronto, Ontario, Canada) discussed that the disappointment regarding the neutral results from SAVOR and EXAMINE resulted from overly high expectations. Dr. Alexandra Kautzky-Willer (Medical University of Vienna, Vienna, Austria) pointed out the positive elements of ongoing CVOTs, applauding CAROLINA (for Lilly/BI’s Tradjenta [linagliptin]) for using an active comparator rather than to placebo and for enrolling patients with newer diabetes as well as more drug-naïve patients – of course, in doing so, they had to agree to two CVOTs, which we believe is well over the top for reasonable FDA expectations, particularly since so many patients take SFUs. We heard Dr. Baptist Gallwitz (Eberhard-Karlz University, Tubingen, Germany) express related sentiments, as he seemed optimistic about CAROLINA’s use of an active comparator and LEADER’s (for Victoza) large sample size. These remarks throughout the conference helped provide direction on how we should plan to interpret the results of ongoing CVOTs and in what ways diabetes researchers can better maximize the value of CVOTs into the future. Indeed, they seem “between a rock and a hard place” and we continue to believe that FDA should bring together a meeting among stakeholders to examine the ongoing value of CVOTs and associated requirements (see our coverage of the recent FDA Public Hearing on interim data disclosure from CVOTs, where speakers from the ADA, Close Concerns, and diaTribe urged the Agency to do just that).

Titles highlighted in blue are new additions that were not mentioned in our daily updates from Vienna, and those highlighted in yellow represent what we felt were the most notable talks of the meeting!

Below is our complete coverage of our coverage pertaining to outcomes trials.

Table of Contents 

Outcomes Trials

Symposium: Hot Topics in DiabetesADVANCE-ON: Post-Trial Observational Study – Study Rationale and Design

Sophia Zoungas, MD, PhD (University of Sydney, Sydney, Australia)

Dr. Sophia Zoungas outlined the design and aims of ADVANCE-ON, a long-term observational follow-up of the ADVANCE trial. ADVANCE was one of the biggest and best-known outcomes trials on glucose lowering (and also blood pressure), having included 11,140 patients with type 2 diabetes. Dr. Zoungas offered a brief summary of the ADVANCE results, which found relative risk reductions in the overall incidence of combined major macrovascular or microvascular events (10%), renal events (11%), and end-stage kidney disease (65%) due to intensive glucose control (consisting of treatment with the sulfonylurea gliclazide and “other therapies as required” to reach an A1c target of 6.5%); no effect was seen on overall mortality or cardiovascular death. The rationale for the ADVANCE-ON study was based on results from the UKPDS follow-up study that found a “legacy effect” of prior intensive control on both microvascular and macrovascular complications. The glucose-lowering component of ADVANCE ended in January 2008, and patient visits for ADVANCE-ON began in 2010 and completed in February 2013. The study’s primary endpoints were mortality and MACE, and secondary endpoints included end-stage kidney disease, death from renal disease, severe diabetes-related eye disease, cardiovascular death, MI, stroke, and major hypoglycemic events.

ADVANCE-ON: Post-Trial Observation Study – Glucose Arm

Sophia Zoungas, MD, PhD (University of Sydney, Sydney, Australia)

Dr. Sophia Zoungas shared that ADVANCE-ON, the long-term observational follow-up of patients in the ADVANCE study that ended in 2013, showed no evidence of a legacy effect of glucose lowering on macrovascular endpoints or even retinopathy. The only major legacy effect was seen on end-stage kidney disease (HR = 0.54; 95% CI: 0.34-0.85). The hazard ratios for major macrovascular events, major clinical microvascular events, and retinal photocoagulation/diabetes-related blindness were 1.00 [0.92-1.08], 0.92 [0.80-1.05], and 0.97 [0.83-1.13], respectively. These results stand in stark contrast to the results from the long-term follow-up of the UKPDS, which found both a microvascular and macrovascular legacy effect. During the subsequent Q&A session, the point was raised that patients in the UKPDS were younger and far more recently diagnosed, whereas patients in ADVANCE were much older and had more longstanding diabetes (average of around eight years diabetes duration at the initiation of the original ADVANCE trial). As Dr. Stephen Gough (University of Oxford, Oxford, UK) pointed out in an excellent talk on outcomes trials on day #1, the window for achieving a benefit on outcomes through glucose lowering likely closes as the disease progresses. Additionally, the study and follow-up for UKPDS (~20 years total) were greater than that for ADVANCE and ADVANCE-ON (~10 years total).

  • In the initial ADVANCE trial, a mean A1c difference of -0.7% was maintained between the groups. This difference disappeared soon after the beginning of the follow-up period, as A1c values stabilizes at a mean of 7.3%. Out of the 10,082 patients that made it to the end of the initial study, 8,494 provided data for the follow-up, and 5,131 made it to the final follow-up visit. The baseline characteristics of patients in the follow-up study were generally comparable to those from the original study, although (due to the healthy survivor effect) there was a reduced proportion of patients with macrovascular and microvascular disease.
  • From the beginning to the end of the follow up, there was an increased use of diabetes medication, indicating an improvement in clinical inertia. There was an increase in the use of metformin and sulfonylureas, and a decrease in other oral agents.
  • For the outcomes results, no subgroup analyses showed any evidence of heterogeneity of effect.

Commentator

Joachim Spranger, MD (Charité-Universitätsmedizin Berlin, Berlin, Germany)

In commentary following the presentations of the ADVANCE-ON data, Dr. Joachim Spranger offered his perspective on the implications of the results. He started with the obvious finding that the study did not provide evidence of a legacy effect of intensive glucose control on mortality or cardiovascular disease, and that the results do not necessitate any change in recommended A1c targets. In his view, these results, taken together with those from the ACCORD trial suggest a need to balance a lower risk of microvascular complications with a possible increased mortality risk when considering stringent glycemic targets for patients. He also argued that the trial provided evidence that treatment with metformin and sulfonylureas (the most common treatment regimen in the study) is safe, as it did not lead to an increased risk of mortality. Dr. Spranger expressed concern about the safety profile of triple therapy with oral medications, which he suggested may have contributed to the excess mortality seen in ACCORD. During Q&A, someone brought up the alternative possibility that harmful effects of sulfonylureas (e.g., hypoglycemia) may have countered the benefits of glucose lowering. Ultimately, we do not believe that ADVANCE or ADVANCE-ON provides sufficient evidence to show a link between specific drug classes and long-term outcomes, as these were studies examining the impact of general intensive glucose lowering. While we think the study is interesting in various respects, we do not even think the results are necessarily generalizable at all to more recently diagnosed people with diabetes.

Panel Discussion

Q: In the glucose trial, what was the A1c target for the standard group?

Dr. Zoungas: In the standard glucose control group, the targets were set by the local teams. We can report on the average effect: we saw an average A1c of 7.3%, so if the target was 6.5%, patients were not achieving it.

Q: Was there any subgroup analysis showing a difference between patients who continued on gliclazide or who shifted to other sulfonylureas or newer drugs like DPP-4 inhibitors?

Dr. Chalmers: We don’t have information on who continued gliclazide. We’ve been very wary – we need to look at all sorts of analyses, but looking at post-randomization analyses is fraught with confounding variables. Patient profiles may be very different than at the start of the study. We certainly haven’t done that analysis, and it’s not one of our highest priorities.

Oral Presentations: Insulin – Clinical Decision Making

People With Type 2 Diabetes With Lower HbA1c Using Insulin Experience Fewer Cardiovascular Events and Death: Results From the CREDIT Study

Nicholas Freemantle, PhD (University College London, London, UK)

Dr. Nick Freemantle presented the full cardiovascular safety results from the Sanofi-sponsored CREDIT study, which was first presented at this year’s ADA. The study was designed to investigate the relationship between glucose control and cardiovascular (CV) adverse events in ~3,000 type 2 diabetes patients beginning insulin in the real world. Compared to landmark CV outcomes trials such as ACCORD and ADVANCE, CREDIT was a observational non-interventional study that did not impact patient or prescriber practice. One-year interim results from CREDIT were presented at EASD in 2011. The full results included four years of follow-up, during which time 147 primary first MACE events occurred. Increased A1c was a predictor for an increased incidence of MACE (HR per 1% increase in A1c: 1.25, p<0.0001) as well as CV death (HR: 1.31, p=0.0027) and stroke (HR: 1.36, p<0.0001) but not myocardial infarction (HR: 1.05, p=0.693) – Dr. Freemantle suggested that the neutrality on myocardial infarction could have been due to chance. Interestingly, there was no correlation seen between severe or symptomatic hypoglycemia and CV death or all-cause mortality.

  • Certainly, more data on the impact of glucose control on CV events is valuable, but given that this was an observational study, it is impossible to know for sure if the increase in CV events observed in the trial was due directly to poorer glucose control or a confounder. A possible next step for the CREDIT investigators is to examine how baseline clinical characteristics differed between patients that achieved better or worse control during the trial, as that data was not included in this presentation.
  • CREDIT was a non-interventional, longitudinal, four-year study conducted across Europe, Asia, and North America. Patients were followed up with no fixed study visit schedule, and hypoglycemia data was collected every year.
  • At baseline: patients had a mean age of 61, mean A1c of 9.3%, mean diabetes duration of nine years, and BMI of 29 kg/m2. Notably, fewer than half of patients were on a statin at the initiation of the trial.

Questions and Answers

Q: You had a very low rate of statin usage in your patients. Might that have had an influence on the results of the study?

A: That isn’t a question that has been raised before, and it had not crossed our mind. We could look to see if there was a difference in the results between those that were on a statin and those that were not.

Q: The baseline A1c at the start of insulin was pretty high, at around 9.3%, so they must have been exposed to severe hyperglycemia for a very long time before the intervention. Is baseline A1c not the more important determinant of risk?

A: I agree that this is a population at a high cardiovascular risk, but this is not significantly different than the broader population, at least in the UK.

Q: You mentioned that one third of the population had a history of macrovascular disease. Did you look at the association between A1c and MACE separately, or look for an interaction with MACE for cardiovascular disease status at baseline?

A: We included status at baseline to account for the increased risk of the population, but we have not tested for any kind of interaction. One reason is that we have limited degrees of freedom to spend on these analyses.

Symposium: Risks and Benefits of New Diabetes Treatments

Relevance of Heart Failure as an Outcome in Diabetes Trials

Hertzel Gerstein, MD (McMaster University, Hamilton, Ontario, Canada)

Dr. Hertzel Gerstein discussed the impact of glucose-lowering drugs (especially DPP-4 inhibitors) on heart failure, and argued that heart failure deserves greater consideration as part of a primary or secondary composite in long-term outcomes trials. He emphasized that hospitalization for heart failure was an exploratory endpoint in SAVOR, and that the 27% increase seen can only be seen as hypothesis generating. There is no clear mechanism to explain a connection between DPP-4 inhibitors and heart failure, and a saxagliptin-specific effect is unlikely. His main take home message is that the medical community will need to wait until other DPP-4 inhibitor CVOTs report – Merck’s TECOS for Januvia (sitagliptin) is expected to end in a few months – this is a long-awaited event. Dr. Gerstein’s voice is part of a rising swell of KOL support for taking heart failure much more seriously as a macrovascular endpoint. Heart failure certainly does have a marked adverse effect on patient quality-of-life and secondary outcomes, but we wonder if adding heart failure or other adverse events to composite outcomes in outcomes trials increases the likelihood of “noise” or chance findings – as we understand it, heart failure can be difficult to adjudicate as consistently as some other CV endpoints.

Questions and Answers

Q: Two weeks ago at the European Society of Cardiology’s Scientific Sessions, a SAVOR analysis was presented showing that severe hypoglycemia in both arms was correlated with increased frequency of hospitalization for heart failure. How would you analyze that?

A: Severe hypoglycemia has been looked at very carefully in many trials, not just SAVOR. We find that severe hypoglycemia predicts a bad outcome for everything: it predicts cancer, death, and fractures. Severe hypoglycemia is just a marker for bad health. People who have severe hypoglycemia are those that are fragile (Editor’s note – we also think many patients who are not fragile are at high risk for hypoglycemia).

Q: Are we entering an era in which we begin stratifying heart failure using BNP or other markers?

A: You’re alluding to a bigger picture: every outcome that we measure as a doctor is in some way a composite outcome. MACE is an obvious composite, but stroke is another composite, as there is ischemic stroke, hemorrhagic stroke, and other forms. There is hospitalized heart failure versus non-hospitalized heart failure, heart failure with NT-proBNP, and other ways to characterize heart failure. The more data we get, the more we can stratify patients. In the end that becomes interesting for experts, but in the end the New York Times is interested in basic hospitalization for heart failure.

Corporate Symposium: Modern Type 2 Diabetes Management – The Experts’ Guide to the Universe of Choices (Sponsored by AstraZeneca)

Preventing Cardiovascular Complications in Type 2 Diabetes: Disappointment or Opportunity?

Steven Gough, MD, PhD (University of Oxford, Oxford, UK)

Dr. Steven Gough’s presentation very eloquently framed many of the issues that have impacted the interpretation of cardiovascular outcomes trials. His central metaphor was that of a closing window: the room for glucose lowering to improve long-term cardiovascular outcomes is likely highest early in the course of the disease, before the burden of calcified atherosclerosis, hypertension, and other risk factors worsen. He encouraged the audience to set realistic expectations regarding ongoing cardiovascular outcomes trials (CVOTs) because they were designed to examine safety and have enrolled patients with higher CV risk and longstanding diabetes, for whom the window for cardioprotection, in his view, “is nearly closed”. He specifically noted that expectations were probably too high for SAVOR and EXAMINE given how short they were. Reading between the lines, it appeared that Dr. Gough was readying the audience for what will likely be a series of neutral results from ongoing CVOTs, encouraging them to not give up hope of real cardioprotection even if the data do not demonstrate a protective effect.

  1. Dr. Gough outlined a “wish list” for future CVOTs:
    1. They will recruit patients soon after diagnosis;
    2. They will recruit patients with minimal established cardiovascular risk, at least compared to current CVOT populations;
    3. They will last for longer than five years;
    4. They will enroll a large study population.
  2. Following Dr. Gough’s presentation, the audience responded to a poll on which glucose-lowering drug class has the highest potential to show a beneficial effect on CV outcomes – GLP-1 agonists came first with 39% of the vote (see below), although Dr. Gough countered that the design of the outcomes study is as important as the specific class being studied. Notably, SFUs got 0%. The full voting results were as follows:
    1. Question: “In your opinion, which of these glucose-lowering agents/classes has the highest potential to show a beneficial effect on CV outcomes?”
      1. GLP-1 receptor agonists: 39%
      2. Metformin: 21%
      3. SGLT-2 inhibitors: 20%
      4. DPP-4 inhibitors: 13%
      5. Insulin: 4%
      6. TZDs: 3%
      7. SFUs: 0%

Panel Discussion

In a short but enlightening Q&A session, Dr. Stephen Gough (who gave the preceding talk) qualified his central argument by noting that although the “window” for cardioprotection may grow narrower as time goes on, it likely never closes completely. Dr. Dan Drucker (Lunenfeld-Tanenbaum Research Institute, Toronto, Ontario, Canada) made the very clinically relevant point that the patients being enrolled in CVOTs are not representative of the broader type 2 diabetes patient population. He pointed out that even when all the currently ongoing CVOTs report their results, clinicians may still have unanswered questions because of the differences in the study populations and their real-world patient populations.

Q: What do you think of the results of our poll?

Dr. Stephen Gough (University of Oxford, Oxford, UK): The most important thing is not the drug class, but the study design. If we want to see a benefit, we will need to use the right study designs. None of these outcomes studies are doing head-to-head comparisons. There are benefits to using glucose-lowering therapy; the most important thing is to use the right drug, at the right time, for the right patient.

Q: You mentioned that glucose lowering may not have much of an impact on cardiovascular outcomes late in the course of type 2 diabetes, but the recent ACCORD results seem to show that intensive control can still improve outcomes.

Dr. Gough: In my slide with the window shutting, I don’t think the window ever completely closed. There is always room for improvements, but the agents we’re talking about can’t do anything about calcified atherosclerosis. An intervention may work in advanced-stage patients, but that is not the ideal population.

Q: Can GLP-1 agonists be cardioprotective?

Dr. Tina Vilsbøll (Gentofte Hospital, Hellerup, Denmark): Overall, what we know today is that there are many pieces of the puzzle that go in the right direction, including lipids, blood pressure, and glucose levels. The one thing drawing in the other direction is the increase in pulse. We will not have the final answer to this until we see the results from the cardiovascular outcomes trials, but from what we have so far, the class at least appears safe.

Dr. Dan Drucker (Lunenfeld-Tanenbaum Research Institute, Toronto, Ontario, Canada): One problem for all of us is that the patients in cardiovascular outcomes trials are not very similar to the patients we generally see in our practice. Even when we have all the results from the outcomes trials, we will still be looking at people in our clinic who don’t look like the people enrolled in those studies, and so those answers will not be definitive.

Corporate Symposium: The Role of Sitagliptin in the Individualized Treatment of the Patient With Type 2 Diabetes (Sponsored by Merck)

Cardiovascular Safety of Antihyperglycemic Agents – What do we Know and What is Still Missing?

Richard Gilbert, MD, PhD (University of Toronto, Toronto, Ontario, Canada)

The discussion on DPP-4 inhibitors and heart failure has not generated any real “new news” of note due to the lack of new data or concrete mechanistic theories. As a result, we were glad that heart failure expert Dr. Robert Gilbert went out of his way to speak frankly on the subject and even provided a theory to explain the signal seen in the SAVOR CVOT. Dr. Gilbert framed the reaction to the SAVOR results as a case of managing expectation: approved drugs such as the TZDs are known to cause heart failure, but the expectations may have been too high for DPP-4 inhibitors due to the signals seen in meta-analyses of early trials that leaned towards cardioprotection. He noted that there are limitations when considering an adverse finding drawn from part of a secondary composite endpoint, and that the broader diabetology community is nowadays much more sensitive to cardiovascular safety issues than it had been previously. Interestingly, Dr. Gilbert compared the time-course of heart failure seen in SAVOR (the signal emerged early then stabilized after six months) to what was seen with retinopathy in the DCCT: a period of early deterioration, followed by stabilization and then (potentially) an improvement. Speculating on a potential mechanism, Dr. Gilbert noted that glucose transporters are down-regulated in the heart during diabetes, and that acute glucose lowering (independent of the drug used) could starve the heart of energy, leading to an increase in heart failure for a period of a few months before the heart could compensate. Although this theory would not explain why the signal has not been seen with all other glucose-lowering agents, overall we thought Dr. Gilbert did an admirable job adding some new commentary to the heart failure debate.

  • Dr. Gilbert dedicated a brief period of time to the TZDs, citing their clear deleterious impact on heart failure. He noted that increases in heart failure are seen even when prior heart failure is an exclusion criterion for a trial, increasing the robustness of the finding. Sulfonylureas all seem to show an increase in heart failure, with no clear differences in the signal between different agents in the class.
  • The lion’s share of the presentation was dedicated to DPP-4 inhibitors and heart failure. Using a rather unique analogy comparing Presidents George W Bush and Barack Obama, Dr. Gilbert suggested that the reaction to the SAVOR and EXAMINE trials were as much a matter of expectations as results. Specifically, the expectations for DPP-4 inhibitors were likely too high, especially given that the SAVOR and EXAMINE CVOTs (for AZ’s Onglyza [saxagliptin] and Takeda’s Nesina [alogliptin], respectively) only achieved modest A1c differences between groups. Dr. Gilbert emphasized that the excess of heart failure seen in SAVOR appeared in the first six months of the trial and dissipated thereafter. He noted that this pattern (early worsening of a signal followed by stabilization and sometimes a long-term benefit) was also seen in previous outcomes trials such as STENO and the DCCT.

Questions and Answers

Chair comment: There has been lots of hoo-hah on heart failure, so thanks for putting it into perspective. Do you think there will be a heart failure signal in TECOS?

A: My prediction would be a slight worsening at first, then stabilization, and then I think there is a chance that we could see protection.

Q: Would you expect a difference between DPP-4 inhibitor and GLP-1 agonists?

A: I think there are issues that have not been clarified in terms of GLP-1 actions to stimulate cAMP secretion and increase heart rate, which is usually considered to have adverse consequences. My apprehension is that they may turn out to be different.

Q: Would you call the SAVOR findings a class effect or specifically for saxagliptin?

A: My answer would be nonspecific for heart failure. We should look at individual trials, and although a meta-analysis has been done on those two trials, we really need more data. My hypothesis is that this is not necessarily an effect that has to do with DPP-4 inhibition, but it may be an acute glucose lowering effect, like what happens with retinopathy.

Q: If you have an individual with a history of cardiovascular disease who is at high risk, would you still add a DPP-4 inhibitor?

A: If it was a patient who had coronary artery disease, then in the absence of heart failure: yes. In the presence of heart failure, especially if it was severe, I would start low, go slow, but I would still use a DPP-4 inhibitor.

Corporate Symposium: Addressing Diabetes Challenges Across the Continuum of Care (Sponsored by Sanofi)

What are we Learning from Cardiovascular Outcomes Trials?

Hertzel Gerstein, MD (McMaster University, Hamilton, Ontario, Canada)

Dr. Hertzel Gerstein spoke strongly in favor of cardiovascular outcomes trials (CVOTs) for diabetes drugs – his message was consistent with what we have heard from him at previous meetings (including last year’s Keystone). He suggested that the medical community is “blessed” to have ~135,000 patients enrolled in CVOTs, suggesting that these trials will provide a wealth of valuable information and will allow providers to be more confident in their treatment recommendations. In particular, Dr. Gerstein defended the value of randomized clinical trials (RCTs) vs. big data: “Don’t be seduced by big data, and don’t let the Kaiser people tell you that you don’t need to do randomized clinical trials” he stated. Observational studies can suggest correlation but cannot prove causation, Dr. Gerstein noted, nor can they uncover unexpected safety signals as reliably as RCTs. This perspective was quite different from what we heard earlier in the day at AstraZeneca’s symposium from Dr. Stephen Gough (University of Oxford, Oxford, UK), who focused on the weaknesses and limitations of CVOTs as they are currently being designed and conducted (too short, population not representative of the broader type 2 diabetes population, etc.).   

  • In a brief discussion on the cardiovascular safety of specific drug classes, we were surprised to hear Dr. Gerstein suggest that sulfonylureas appear to be safe and could possibly be cardioprotective. This suggestion runs counter to what we have heard from most other conference speakers – although there is no rigorous RCT data on SFUs’ CV safety, epidemiological data so far appears to suggest (if anything) an increase in risk. 

Corporate Symposium: A Special Report on Type 2 Diabetes – Finding the Right Treatment Routes to Optimize Your Patients’ Journey (Sponsored by Lilly/BI)

This Just In: CV Outcome Trials

Alexandra Kautzky-Willer, MD (Medical University of Vienna, Vienna, Austria)

Dr. Alexandra Kautzky-Willer’s review of ongoing cardiovascular outcomes trials focused primarily on the trials investigating Lilly/BI compounds: CARMELINA and CAROLINA for Tradjenta (linagliptin) as well as EMPA-REG OUTCOME for the newly approved Jardiance (empagliflozin). She lauded CAROLINA for addressing many of the limitations of other CVOTs: it is a head-to-head trial vs. an active comparator (glimepiride) rather than placebo; it is likely to run for fairly long (six to seven years); and it has enrolled patients with newer diabetes as well as more drug-naïve patients. Dr. Kautzky-Willer concluded that CAROLINA will provide information that will be highly relevant and useful for providers and patients that are selecting a second-line therapy after metformin. Moving from DPP-4 inhibitors to SGLT-2 inhibitors, Dr. Kautzky-Willer highlighted that EMPA-REG OUTCOME will be the first CVOT to report for the class (we learned last year that the estimated completion date for the trial was moved up by three years, to 2015). She also suggested that the trial has the potential to show cardioprotection due to SGLT-2 inhibitors multiple beneficial effects on CV risk factors.

-- by Melissa An, Adam Brown, Varun Iyengar, Emily Regier, Manu Venkat, and Kelly Close

-- The authors thank Eric Chang, Hannah Deming, Jessica Dong, Nina Ran, and Melissa Tjota for additional help on conference writing and editing