Digging into the TULIP Trials and AstraZeneca’s Blossoming Ambitions in Lupus

After nearly 15 years of studying a novel pathway to treat immune-driven diseases, AstraZeneca surprised the medical community when it announced that the first of its two Phase III trials in systemic lupus erythematosus (SLE) did not meet its primary endpoint. But instead of unrooting years of progress, the company turned its second trial into a success through applied learnings, engagement with the lupus community, and arming trial sites with the right tools.

Moe Alsumidaie: After TULIP-1 missed its primary endpoint, how did the anifrolumab program evolve into a success with TULIP-2?

Micki Hultquist: Before we dive in, let me set the stage for you by backing up and discussing lupus. Lupus is an incredibly debilitating disease for patients. There has only been one new medicine approved in the last 60 years, and the medicines that these patients often rely on, such as oral corticosteroids or immunosuppressants, also contribute to the comorbidities they experience – so there is a desperate need for alternative treatments.

At AstraZeneca, we spent the last 15 years focused on targeting the interferon pathway, and we’ve been encouraged by our data. I led the team through the Phase II results in SLE, and we were anticipating hitting this first Phase III study out of the park – so the unblinding of TULIP-1 surprised us. Afterward, somebody suggested that we consider delaying the unblinding of TULIP-2, which had not been read out yet. And I’ll be honest, my initial instinct was a concern about staying to our timelines, but then we got the final tables from the Contract Research Organization (CRO), and it became clear that the right decision was to pause, apply learnings from the encouraging results seen in TULIP-1 across other endpoints, and adjust our approach to TULIP-2.

So that’s what we did, and after unblinding TULIP-2 about a year later, we saw that the primary endpoint was met, and the data were highly consistent with the encouraging results from TULIP-1 – that anifrolumab resulted in early sustained response using the British Isles Lupus Assessment Group-Based Composite Lupus Assessment (BICLA; an incredibly stringent disease activity endpoint), sustained steroid reduction and improvements in skin disease manifestations.

MA: What exactly were the issues from TULIP-1 that you uncovered, and how did you address those issues in TULIP-2?

MH: While we failed the primary endpoint of patient response using the SLE Responder Index-4 (SRI-4), this is one of two endpoints in lupus clinical trials that are typically considered to measure disease activity – the other being BICLA, for which anifrolumab demonstrated compelling results. So, it was thinking through the new data we had on anifrolumab from TULIP-1 and realizing BICLA was the right primary endpoint for TULIP-2. We also pulled in lupus experts to align on the path forward for TULIP-2 to make sure everybody was thinking through the solution.

In TULIP-1, the SRI-4 endpoint was not met; however, we saw a reasonably strong BICLA response with a difference versus placebo of about 16%. That didn’t happen in TULIP-2—we saw pretty consistent results between SRI-4 and BICLA measurements, with about an 18% difference in SRI-4 response and 16% difference in BICLA response.

MA: So, if I understand correctly, it was the assessment of these endpoints that made the difference in TULIP-2?

MH: We don’t fully understand what drove the difference in TULIP-1 between the SRI-4 and the BICLA responses. We are continuing to explore and hope to release a publication with our findings in due course.

Lupus clinical trials are challenging because lupus impacts any organ system, and the endpoints that we use to measure improvements are looking at every organ in the body and assigning some mathematical algorithm to it, which is complicated.

And as you know, there have been multiple failed new medicines in lupus. So, it’s a graveyard of clinical products. And my opinion is that it’s mainly due to these endpoints that are trying to mimic what’s going on in the body, which is almost impossible to do. Lupus presents differently in every single person, so you can’t use the same measure to show real-life improvements in these patients. It’s not like rheumatoid arthritis, where you’re always looking at joints. You can’t do that here.

MA: So, would you say this unpredictable way of measurement is due to the nature of the disease rather than sites that aren’t trained on these measurements?

MH: I think that’s a fair assessment. Of course, it is essential to train sites and ensure there are site investigators who are well-trained in lupus and lupus endpoints. The endpoints used in clinical trials are not typically used in practice, so the investigators must know the nuances of using these measurements.

MA: Compared to your experiences with TULIP-1, what sorts of operational challenges did you experience when running the TULIP-2 trial?

MH: We started the TULIP-1 and TULIP-2 trials almost concurrently, and I think many of the operational challenges we faced stemmed from the transition from Phase II to Phase III, as in any clinical trial program. This meant increasing our participation and global footprint, and then also working with a CRO required an extra layer of communication.

The AstraZeneca team attended site initiation visits to present the Phase II data and was involved in the oversight of the study, so our approach was hands-on from the beginning. In addition, we worked to ensure strong relationships with our investigators and the CRO, so we worked as one anifrolumab team to help prevent communication or operational challenges.

MA: Can you tell us about any patient-centric initiatives you carried out when designing the trials, such as work with advocacy groups or dry runs with patients?

MH: We did align ourselves with advocacy groups and conducted small patient focus groups, even a separate almost “pilot trial” where they did some screening visits with patients. Patient centricity is a critical component of study design in our portfolio—it’s one of AstraZeneca’s core values. My team meetings often include discussions about patients and how central they are in our profession. It’s vital to me.

In particular, in the TULIP program, we sought patient feedback on the study design to ensure their voice was included in the process. Again, we can see this by only performing assessments vital to the study’s objectives through our attempt to alleviate the burden on patients.

MA: Could you tell me a bit about any challenges you faced with recruitment or enrollment?

MH: We found that the best source of success is diligent site selection and aligning ourselves with experienced sites. The best sites are those that understand lupus endpoints and have experience in lupus clinical trials and sites that already had relationships with the participants.

We also found some success establishing a site and study coordinator network so that sites with high enrollment could share their best practices and help others.

MA: And lastly, what systems did you implement to reduce data variability and improve the data quality of these disease measurements?

MH: I recommend routine and diligent training of the investigators—that’s critical. We also utilized an adjudication committee that reviewed patient entry and the core data to ensure responses were internally consistent and watchable. That, I think, was important and carried through from Phase II into both TULIP-1 and TULIP-2.

MA: And what happens if there is a disagreement between the adjudicator and the investigator?

MH: A process that remediates it involves a lot of back and forth between the adjudicator and the site until they can come to a solution.

Moe Alsumidaie, MBA, MSF, is a thought leader and expert in the application of business analytics toward clinical trials, and Editorial Advisory Board member for and regular contributor to Applied Clinical Trials.