The value of AMH in assessing ovarian function and damage in

women with cancer

Richard A AndersonElsie Inglis Professor of Clinical Reproductive Science

University of Edinburgh

Chemotherapy: immediate and late effects on the ovary

• Depletion of growing folliclesHimelstein-Braw R, Peters H and Faber M (1978)

Morphological study of the ovaries of leukaemic children.

Br J Cancer 38, 82-87

• Premature ovarian failureChapman RM, Sutcliffe SB and Malpas JS (1979)

Cytotoxic-induced ovarian failure in women with Hodgkin's disease. I. Hormone function.

JAMA 242, 1877-1881

Effects of cancer therapy on the ovary

Cancer

Bloodvessels

PrimordialFollicles AMH

Estrogendeficiency

Infertility

Potentialfertility/subfertility

Treatment

GrowingFollicles

Post treatmentamenorrhoea

Recovery, of variable duration

Premature ovarianinsufficiency

AMH

AMH

Jayasinghe, Wallace and Anderson 2018 Expt Rev Endo Metab

Biomarkers: AMH, AFC, mensesClinical outcomes: fertility, age at menopause

Letourneau et al 2012 Cancer 118, 1710

Pregnancy: HR 0.87 (0.81-0.94)

Alkylators only at highest doseBusulfan and Lomustine

Live birth to female childhood cancer survivors: chemo only

Chow et al Lancet Oncol 2016

Hazard ratio for menopause <40 yrs in Hogdkin Lymphoma

0 5 10 15 20 25 30 35 40

No alkylating, no pelvic RT

Pelvic RT

Alkylating, no pelvic RT

Alkylating, pelvic RT

ABVD

ABVD with pelvic RT

Swerdlow AJ et al 2014, J Natl Cancer Inst

All adjusted for age, overall n=2127 (though data only from 50%)

Age-related changes in the ovarian reserve

Wallace and Kelsey 2010 PLoS One 5; e8772

Which stages of follicle growth are key targets of cancer therapies?

Loss of growing follicles may increase growth activation?

Focal cortical fibrosis in ovaries exposed to chemotherapy

Meirow D et al. Hum. Reprod. 2007;22:1626-1633

The ovarian stroma and vasculature are also targets

Normal controlGreen: Masson stain for collagenAfter chemotherapy

Prominent thickening and hyalinization, with narrowing

/obliteration of the lumen

Effects of cancer therapy on the ovary

Cancer

Bloodvessels

PrimordialFollicles AMH

Estrogendeficiency

Infertility

Potentialfertility/subfertility

Treatment

GrowingFollicles

Post treatmentamenorrhoea

Recovery, of variable duration

Premature ovarianinsufficiency

AMH

AMH

Jayasinghe, Wallace and Anderson 2018 Expt Rev Endo Metab

Biomarkers: AMH, AFC, mensesClinical outcomes: fertility, age at menopause

The variability in ovarian activity after cancer treatment

Ovari

an

Activity

CancerDiagnosis

Onset ofdisease

Age-relatedDecline

TreatmentTime

POIThreshold

Recovery Period

Key variables: age and treatment

Jayasinghe, Wallace and Anderson 2018 Expt Rev Endo Metab

AMH reflects the number of small growing follicles

Inhibin B, estradiol

AMH

Anderson RA 2012 Clin Endocrinol 77, 652

>60% from 3-8 mm antral follices

Jepperson, Anderson et al 2013 MHR 19, 519

Altered AMH to primordial relationship:IllnessAgeTreatment

AMH reveals ovarian damage in childhood cancer survivors despite regular cycles

Controls

Cancer survivors

Inhibin B

0

100

200pg/ml

FSH

0

4

8

12

IU/L

E2

0

100

200

pmol/L pmol/L

AMH

0

10

20

30

*

Bath LE et al 2003 Human Reprod 18 2368

AC 3CMF 3A-CMF 7E-CMF 18 FECT-T (TACT) 9TANGO 2

Post chemoTamoxifen 26 Goserelin + Tam 8Arom inhib 11None 4

6-9 months

Recruited n=56

No chemoGoserelin + Tam 8Tamoxifen 5Gos + anastrozole 1

Chemotherapy 42

Surgery

‘Late’ analysis35 at 4 years33 at 5 years (79%)(recurrence, TAH/BSO)

USS:27 pretreatment21 at 5 years

Effect of chemotherapy in eBC:acute toxicity and long-term prediction

Analyse ovarian activity here

In relation to predictive markers here

Anderson RA et al 2006 Human Reprod 21, 2583

Effect of chemotherapy in eBC:acute toxicity

0

10

20

30

40

50

0 3 6 9 12

FSH and LHIU/L

Months Months

0

200

400

600

800

0 3 6 9 12

Estradiolpmol/L

0

20

40

60

80

0 3 6 9 12

Inhibin Bng/ml

0.0

0.4

0.8

1.2

0 3 6 9 12

AMH

ng/ml

n=42

Anderson RA et al 2006 Human Reprod 21, 2583

Effect of chemotherapy in eBC:long-term prediction

*

Inhibin B

0

20

40

60

80

100

CRA Menses

AMH

0.0

1.0

2.0

3.0

4.0

CRA Menses

E2

0

100

200

300

CRA Menses

Anderson and Cameron 2011 JCE&M 96, 1336

AMH at diagnosis of early breast cancer is higher in those women who will still be having menses 5 years later

0

10

20

30

40

50

60

70

80

90

100

0 20 40 60 80 100

AMH

Age

Prediction of amen

AMH AUC 0.91 (0.870-1.01)

Pretreatment hormones vs late ovarian function

Clinical application:predictive mosaic chart in eBC

Anderson et al 2013 Eur J Cancer

sensitivity 98.2% specificity 80.0%for correct classification of amenorrhoea

n=75

Pretreatment AMH is reduced in girls with childhood cancer

van Dorp W et al. Hum. Reprod. 2014;29:337-342

n=208Temperature and CRP inversely correlated with AMHHb positively correlated

AMH profiles after chemotherapy

Are AMH levels here discriminatory?

Is AMH a good diagnostic here?

Leonard et al 2017 Ann Oncol

227 women with breast cancer, randomised to ± goserelin during chemotherapy

AMH as a diagnostic test in POI?

Contr

ol

WHO 1

WHO 2

(PCOS)

WHO

2 (a

ll)

WHO 3

HPR

L

0

25

50

75

100175

200 ** **

*

DIAGNOSIS

Se

rum

AM

H (p

mo

l/l)

Li et al 2011 Fertil Steril 96, 774

• Not part of the diagnosis at present

• Will increased assay sensitivity help?

• Useful in ‘fluctuant’ stage of condition when E2 and FSH very variable?

• But value post chemo excluded in STRAW+10

0

50

100

150

200

E2 p

mol/l

Baseline EoT 12 mo 24 mo

0

20

40

60

80

100

FS

H (

IU/L

)

Baseline EoT 12 mo 24 mo

0

1

2

3

4

6

8

10

AM

H (

pm

ol/l

)

Baseline EoT 12 mo 24 mo

Serum FSH and AMH by POI at 24 months. Data from all women from OPTION

Roche automated AMH assay

Red, not POI

Blue: POI (amenorrhoea plus FSH >25IL/L).

N=96 and 28 respectively; median ± 95% confidence intervals.

Can AMH diagnose POI after chemo?

ROC AUC of 0.86

sensitivity 100%

specificity 73%, LR 3.7

So YES at 24 months

Anderson et al 2017 Eur J Cancer

Importance of age for recovery of ovarian function after chemotherapy

0

1

2

3

4

5

AM

H (

pm

ol/l

)

EoT 12 mo 24 mo

0

20

40

60

80

FS

H (

IU/L

)

EoT 12 mo 24 mo0

50

100

150

200

250

Estr

adio

l (pm

ol/l

)

EoT 12 mo 24 mo

Women aged ≤ 40 (purple) vs >40 years (orange)

n=62 and 81, median ± 95% CI.

Anderson RA et al 2017 Eur J Cancer

Data from

OPTION trial

EOT predictive analysis in women >40 yrs

0 25 50 75 1000

25

50

75

100

100% - Specificity%

Sensiti

vity%

0 25 50 75 1000

25

50

75

100

AMH and FSH at end of treatment for

prediction of POI at 24 months

OPTION control group, n=32 months

undetectable detectableAMH at

EOT

E2

@12

mo

AM

H-v

e

E2

@24

mo AM

H-v

e

E2

@12

AM

H+

E2

@24

AM

H+

0

500

1000

1500

100

Estr

adio

l pm

ol/l

12 12 2424

AMH AUC 0.89

sensitivity 91%, specificity 82%

FSH AUC 0.77

sensitivity 100%, specificity 55%

Anderson RA et al 2017 Eur J Cancer

AMH and ovarian function after chemotherapy

AMH for• Pre chemo prediction• Post chemo diagnosis of POI• Immediate post chemo prediction?

The AMH normal rangefrom birth to menopause

Kelsey et al 2011 PLoS One 6: e22024

0 10 403020 50

AMHng/ml

Key featuresDetectable in girls of all agesRise through childhoodPeak at 24 yrs

Pre

Cyc

le 1

Cyc

le 2

Cyc

le 3

Cyc

le 4

Cyc

le 5

Cyc

le 6

0.0

0.5

1.0

1.5

2.0

2.5

****

*** ***

AM

H (n

g/m

l)

AMH: application in childhood cancer

Pre End Recovery0

1

2

3

* **

AM

H (n

g/m

l)

High risk

Pre End Recovery0

1

2

3**

AM

H (n

g/m

l)

Medium/low risk

22 girls age 0.3-15yr17 prepubertal

Brougham et al 2012 JCE&M 97, 2059

AMH in 3 girls with cancer

0 50 100 1500.0

0.5

1.0

1.5

2.0

Weeks

Age 14.6: Hodgkin’s lymphoma

Brougham et al 2012 JCE&M 97, 2059

0 25 50 750.0

0.2

0.4

0.6

0.8

1.0

Weeks

AM

H (n

g/m

l)

Age 1.2; neuroblastoma

Pre End Recovery0

1

2

3

* **

AM

H (n

g/m

l)

0 50 100 150 2000.0

1.0

2.0

3.0

Age 2.4; rhabdomyosarcoma

Pre End Recovery0

1

2

3**

AM

H (n

g/m

l)

How predictive is this?

Sowers MR et al. J Clin Endocrinol Metab 2008;93:3478-3483

Does AMH predict natural menopause?

50 women followed prospectively(Michigan Bone Health and Metabolism Study)6 annual assessments

Mean initial age 42 yr

AMH related to both time to and age at FMP

Inhibin B less predictive of both

0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0 1 2 3 4 5 6

Cum

ula

tive

pro

po

rtio

n o

f w

om

en

ach

ievin

g p

regn

an

cy

Time from cessation of birth control (cycles)

0

10

20

30

40

50

60

70

19 21 23 25 27 29 31 33 35

AM

H (

pm

ol/L

)

Age (years)

AMH and fecundability

Hagen et al 2012 Fertil Steril

AMH quintiles, middle 3 combined

981 women aged 30 to 44, trying to conceived max 3 months at study entry

Steiner AZ et al, 2017, JAMA

Adjusted for age, smoking, contraception, BMI, race, prev pregnancy

AMH and fertility in older women

What about low toxicity regimens?RATHL trial in Hodgkin Lymphoma

2 cycles ABVD

Full dose, on schedule

PET 2 -vePET 2 +ve

4 cycles ABVD

PET2

PET 1(Staging)Stage II (adverse),III,IV,IPS 0-7

Over 18PS 0-3

Randomise

4 cycles AVD

Follow-up (no RT)

4 cycles BEACOPP-14

or 3 eBEACOPP

PET3

PET 3 -vePET 3 +ve

RT or salvageregimen

2 cycles BEACOPP-14 or 1 eBEACOPP

No RT

Ovarian substudy method

Women aged 18-45 were recruited(ethics approval/consent)

Blood samples:• Pre-treatment• After 2 cycles ABVD• End of chemo• 1, 2, 3 years later

• Analysed for AMH, FSH (Roche)

Johnson P et al. Adapted Treatment Guided by Interim PET-CT Scan in Advanced Hodgkin's Lymphoma.N Engl J Med. 2016; 374: 2419-29

Pre

treat

men

t

2 cy

cles

End

of t

reat

men

t1

yr2

yr3

yr0.01

0.1

1

10

100

AM

H (

pm

ol/L

) Blue: ABVDRed: BEACOPP(after 2 cycles of ABVD)

Effects of A(B)VD and BEACOPP on ovarian function

Anderson RA et al 2018 Lancet Oncol

Hazard ratio for menopause <40 yrsin treatment of HL

0 5 10 15 20 25 30 35 40

No alkylating, no pelvic RT

Pelvic RT

Alkylating, no pelvic RT

Alkylating, pelvic RT

ABVD

ABVD with pelvic RT

Swerdlow AJ et al 2014, J Natl Cancer InstAll adjusted for age, overall n=2127 (though data only from 50%)

Main relationships: AMH, age, recovery

0

20

40

60

80

100

15 20 25 30 35 40 45 50

AM

H p

retr

ea

tme

nt

Age (years)

AMH pretreatment vs age

0

20

40

60

80

100

0 20 40 60 80

AM

H a

t 2

ye

ars

AMH pretreatment

AMH pretreatment vs 2 yr levels

Spearman r=0.71, p=0.0002slope =1.05Overall, AMH at recovery reflects pretreatment level

Anderson RA et al 2018 Lancet Oncol

Is AMH recovery always good?

0

100

200

300

400

500

15 20 25 30 35 40 45 50

AM

H r

eco

ve

ry (

%)

Age (years)

AMH recovery by age

r=-0.48, p=0.001

Older women show reduced recovery

0

100

200

300

400

500

0 20 40 60 80

AM

H r

eco

ve

ry (

%)

AMH pretreatment

AMH recovery by pretreatment AMH

r=-0.02, p=0.9

Women with low AMH show full recovery

Anderson RA et al 2018 Lancet Oncol

Confirmation of impact of age on recovery

0

20

40

60

80

100

120

140

160

Age <35 Age 35+

% r

eco

very

P<0.0001

By age

Multiple linear regression analysis vs AMH recovery:age (beta -0.43, p=0.004) pretreatment AMH (beta -0.15, p=0.3)

0

20

40

60

80

100

120

140

1 2Above Below

By median AMH (9.8 pmol/l)

ns

Anderson RA et al 2018 Lancet Oncol

Different to breast cancer data: older population, more toxic treatment

FSH recovery after A(B)VD is also dependent on age

Anderson RA et al 2018 Lancet Oncol

recovery to <25IU/L

Pre

trea

tmen

t

EOT

1 ye

ar

2 ye

ars

0

25

50

75

FS

H (

IU/L

)

FSH: effect of age with ABVD treatment

ABVD <35

ABVD 35+

*

*

* *

Age <35Age 35+

% recovered Age <35 35+

At 1 year 83% (77 – 88) 54% (43 – 66)

At 2 years 96% (93 – 98) 83% (73 – 91)

At 3 years 98% (95-99) 93% (85-97)

Mean±sem

Conclusions

Need for accurate, patient-specific risk to fertility and ovarian function

Extrinsic issues: proposed treatmentIntrinsic issues: age and ovarian reserve

Rational and effective use of FP techniquesTruly informed consent for interventions

AMH is a valuable index of ovarian function in oncology patientsPretreatment assessmentPost treatment: immediate, and long term

Key collaborators and funding

Hamish Wallace, and colleagues, Paediatric oncology, Edinburgh

David Cameron and colleagues, Edinburgh Breast Unit

Bob Leonard and OPTION investigators

Peter Johnson and RATHL investigators

Tom Kelsey, Mathematician, St Andrews University

Beckman Coulter and Roche Diagnostics for assay reagents