1. News at the horizon Concepts of clinical behaviour in the ILDs
Athol Wells
Royal Brompton Hospital

2. AUW has received consultancy or lecturing fees from Actelion, Bayer, Boehringer Ingelheim, Gilead, Intermune/Roche

3. CHP, our most difficult disease
The importance of observed or anticipated disease behaviour

4. Patterns of disease behaviour
Depending upon definitions and triggers, we have over 1000 individual ILDs
However, this does individualise separate HP and drug triggers. In reality, we have perhaps 50 ILDs
But with regard to disease behaviour, logical treatment goals and monitoring, we have only five disorders

5. HP versus CHP CLINICAL BEHAVIOR TREATMENT GOAL
02/09/12
HP versus CHP
CLINICAL BEHAVIOR
TREATMENT GOAL
MONITORING STRATEGY
Reversible & self-limited
Remove possible cause
Short term (3-6 month) observation to confirm disease regression
Reversible disease with risk of progression
Initial response & then rationalize longer term therapy
Short term observation to confirm treatment response.
Long term observation to ensure that gains are preserved
Stable with residual disease
Maintain status
Long term observation to assess disease course
Progressive, irreversible disease with potential for stabilization
To prevent progression
Long term observation to assess disease course
Progressive, irreversible disease despite therapy
To slow progression
Long-term observation to assess disease course, need for transplantation and effective palliation
Travis WD et al. Am J Respir Crit Care Med 2013; 188:733-48

9. Can we rely on the evidence base to treat CHP with confidence?
In current cohorts, disease behaviour is not taken into account in management

10. Treated total, n=93
Prednisolone, n=41 Pred + Aza, n=24 Pred + MMF, n=28

11. Caveats
Neither these data nor any other evaluate treatment versus no treatment in a robust fashion
Those receiving therapy had disease that was much more severe than those untreated and had worse outcome.
The worthwhile data in this study relate to
Adverse events in the shorter term, compared between regimens in the primary cohort: high dose prednisolone regimens strikingly more adverse events
Differences in long term outcome (FVC change, mortality) across five cohorts: zilch!!!

13. Total, n=70 MMF, n=51 Aza, n=19 Pred 12mg daily

14. Why the evidence base does not mirror the real world…..
These cohorts are reported as a “one size fits all” treatment approach
In reality, we make a disease behaviour judgement whether a major response is plausible and if so, it is important not to miss an opportunity
If disease is irreversible, why persist with high dose steroid therapy? We should be finding a civilised long term approach which, we hope, will prevent or retard disease progression.

15. CLINICAL BEHAVIOR TREATMENT GOAL MONITORING STRATEGY
Reversible & self-limited
Remove possible cause
Short term (3-6 month) observation to confirm disease regression
Reversible disease with risk of progression
Initial response & then rationalize longer term therapy
Short term observation to confirm Rx response. Long term observation to ensure that gains are preserved
Stable with residual disease
Maintain status
Long term observation to assess disease course
Progressive, irreversible disease with potential for stabilization
To prevent progression
Progressive, irreversible disease despite therapy
To slow progression
Long-term observation to assess disease course, need for transplant or effective palliation

16. This is where we are now positioned. News on the horizon:
The future shape of IPF pathogenesis: precision medicine versus the progressive fibrotic phenotype
The use of disease behaviour to modify diagnosis
The future development of the disease behaviour classification

17. Where lies the future of “IPF”
Will pathogenetic thinking be driven by “splitters”, who define IPF sub-groups from observed differences, in the hope of precision medicine?
Will it be driven by “lumpers” (who search for a big picture over and above diagnostic nuance)? Will the amalgamation of IPF patients and non-IPF patients with IPF-like disease behaviour be more fruitful?
To argue for the primacy of either assumes a level of knowledge that we do not yet possess

18. These are not mutually exclusive
We need smarter splitting and smarter lumping.
IPF should continue to be sub-grouped. Biomarkers associated with IPF and with more progressive IPF should be explored in progressive non-IPF disorders
Examples of shared pathogenetic pathways between IPF and non-IPF fibrotic disease are now emerging
We are not permitted to use anti-fibrotic therapy in progressive fibrotic non-IPF disorders

19. A great deal now depends upon current anti-fibrotic trials in grouped non-IPF disorders
If these studies are positive, we will see the dawn of classification across ILD by disease behaviour

20. The “progressive fibrotic phenotype” is here to stay

21. Can the progressive fibrotic phenotype be identified with confidence at baseline?
This would be ideal, especially if anti-fibrotic trials in non-IPF are positive
There are a number of studies in which biopsy or HRCT features (usually of UIP) are associated with a bad outcome in CHP, CTD-ILD, unclassifiable disease
But is baseline data really good enough?

22. Study of outcome in RA-ILD
157 patients with RA-ILD (Edinborough, London)
Examination of a priori variables against mortality
HRCT pattern (UIP, probable UIP, inconsistent with UIP)
Staging of disease severity using the Goh SSc-ILD criteria
Jacob J et al: work in progress

23. HRCT patterns Appearances typical of IPF
Peripheral honeycombing, distribution atypical for IPF
Probable UIP
Probable UIP, distribution atypical for IPF
Inconsistent with UIP

24. IPF (n=284) blue – mean survival 2.9 years
RAILD definite UIP pattern (n=55) green - mean survival 3.5 years
RAILD probable UIP pattern (n=56) yellow - mean survival 4.6 years
RAILD inconsistent (n=46) violet – mean survival 5.3 years
Jacob J et al: work in progress

25. The clinical need for staging
Decisions are dichotomous: treat or not, enrol in treatment trial or not
We need definition of high and low risk disease
We need to STAGE lung disease in CTD-ILD

26. Staging by severity: what is needed
Simple, user-friendly system
Accurate depiction of high and low risk in moderately experienced hands
Conceptually easy

27. Disease extent determines mortality
100 80 60 40 20
120
Limited
Duration of follow-up (months)
Extensive
Survival (%)
Goh NS et al. Am J Respir Crit Care Med. 2008; 177:

28. The UKRSA staging system
HRCT extent
>20%
<20%
Indeterminate
FVC >70%
FVC <70%
Extensive Disease
Mild Disease

29. IPF (n=284) blue
Extensive RAILD (n=68) green
Limited RAILD (n=88) yellow
Jacob J et al: work in progress

30. Separate UIP outcome subsets
Non-RA-UIP, n = 18
Idio. NSIP, n = 66
CTD-NSIP, n=57
RA-UIP, n=19
IPF-UIP, n=203
Park JH et al. Am J Respir Crit Care Med 2007; 175:

31. Can the PFF be identified with confidence at baseline?
Not yet!!!
Based on current data, the PFF may eventually be defined from baseline HRCT (with biopsy findings in some cases) and emerging molecular data
But for the moment, the PFF equates to progression despite treatment considered to be appropriate in individual ILDs

32. Disorders in which observed FVC (or HRCT) progression drives mortality
Idiopathic NSIP
SSc-ILD
CHP
RA-ILD
Unclassifiable lung disease
Latsi PI et al. Am J Respir Crit Care Med 2003; 168:531-7
Jegal Y et al. Am J Respir Crit Care Med 2005; 171:639-44
Goh NS et al. Arthritis Rheumatol 2017; 69:1670-8
Gimenez A et al. Thorax 2017; epub ahead of print
Solomon JJ et al. Eur Respir J 2016; 477:588-96
Jacob J et al. Respir Med 2017; 130:43-51

33. Disease behaviour in differential diagnosis
The difficult differentials with regard to management (anti-fibrotic therapy versus immunomodulation):
IPF versus fibrotic NSIP
IPF versus CHP

34. How likely is IPF: PE probabilities…
Definite ( %)
Probable ( %)
Indeterminate ( %) - management????
Possible ( %)
Unlikely (0 – 5%)

35. A highly reliable stress test: are you in need of a holiday
A highly reliable stress test: are you in need of a holiday? Examine the next slide closely……..

37. For management purposes, a diagnosis can be…..
Established >90%
Provisional - high confidence %
- low confidence %
Unclassifiable <50%
Ryerson CJ et al. Am J Respir Crit Care Med 2017: 196:

39. “If you don’t know where you are going, any road will get you there”

40. Multidisciplinary diagnosis is very like the Cheshire cat
Consider proposed categories for confidence of diagnosis

41. An established diagnosis

42. A provisional diagnosis

43. Unclassifiable disease

44. If the likelihood of IPF and fibrotic NSIP are both roughly 50% and you treat as for NSIP, how strongly should the treated course influence diagnostic likelihood?
How likely is IPF if disease progresses despite treatment over one year?
We can model indeterminate disease (likelihood 35% - 65%)

45. Longitudinal behaviour in fibrotic IIP
Biopsy-proven IPF (n=84) or idiopathic fibrotic NSIP (n=72)
Treated with corticosteroids or combined steroid/ immunosuppressive drugs
At 12 months, FVC decline (5% or greater):
IPF %
NSIP %
Zappala CZ et al. Eur Respir J 2010; 35:830-6

46. If IPF likelihood at baseline = 65%
If disease progresses, IPF:NSIP = 55%:24%
This likelihood of IPF is computed by multiplying the baseline likelihood (LR = 65/35) by the progression likelihood (LR = 55/24) …….. LR of IPF now 117/28 = 117/145 = 81%
This shifts the IPF diagnosis from “provisional low confidence” to “provisional with high confidence”

47. If IPF likelihood at baseline = 35%
If disease progresses, likelihood of IPF = 55%. This provides a provisional multidisciplinary diagnosis of IPF and opens the door for anti-fibrotic therapy
If disease is stable, in the indeterminate 35%-65% range, likelihood of NSIP is 48-75%
At 50/50 likelihood at baseline, 70% likelihood of IPF if disease progresses, 63% likelihood of NSIP if disease is stable
When diagnostic likelihoods at baseline are a close call, the treated course provides a provisional multidisciplinary diagnosis

48. The disease behaviour classification should be developed across ILD and not confined to unclassifiable disease
Future thinking in IPF (the progressive fibrotic phenotype) and the use of disease behaviour to refine MD diagnosis justify this initiative
And there are other advantages……..

49. We need this approach in order to communicate

50. CTD-ILD: a poll (n=350). With regard to clinical utility, the IIP classification is:
Intuitive, user friendly and readily understood by patients
Is somewhat helpful to clinicians and patients although not always straightforward
Difficult to grasp, arduous to apply but worthwhile
Opaque, poorly understood by patients and of little clinical value

51. CTD-ILD: a poll (n=350). With regard to clinical utility, the IIP classification is:
Intuitive, user friendly and readily understood by patients
Is somewhat helpful to clinicians and patients although not always straightforward
Difficult to grasp, arduous to apply but worthwhile
Opaque, poorly understood by patients and of little clinical value 9%
17%
30%
43%

52. If correct in principle, this approach should apply generally
02/09/12
If correct in principle, this approach should apply generally
NSIP HP
CTD-ILD
Sarcoidosis
Drug-induced lung disease

53. Summary: key concepts
We are entering an era in which ILD classification will become more dynamic
In suspected IPF, the concept of the progressive fibrotic phenotype will increasingly inform pathogenetic thinking and management
Studies of disease behaviour will allow greater precision in multidisciplinary diagnosis
The disease behaviour classification might usefully be applied across ILD